threads / patch / 64222

patch, 49 partsrepack: prepare for incremental MIDX-based repacking

Subject: [PATCH 00/49] repack: prepare for incremental MIDX-based repacking

## tl;dr

163 messages between Sep 28, 2025 and Oct 17, 2025. Diffs are folded; open one to read it.

replies: 162people: 4as markdown or json

Taylor Blau· Sep 28, 2025, 22:07 UTC · lore
Note(s) to the maintainer:
 * This series depends on ps/packfile-store at dd52a29b78 (packfile:
   refactor `get_packed_git_mru()` to work on packfile store,
   2025-09-23).
 * This series is split out of my local copy of
   tb/incremental-midx-part-3. I suggest queueing it as
   tb/incremental-midx-part-3.1 ;-).

When developing an incremental MIDX/bitmap-based repacking strategy, I thought that it might be useful to clean up builtin/repack.c before adding any amount of new functionality.

At more than 1,700+ lines of code, the 'repack' builtin has grown quite unwieldy over the years. It has quite a bit of repack-related functionality, all implemented within the builtin itself. It knows how to: manage cruft packs, perform geometric repacking, write new MIDXs, repack promisor objects, keep track of existing packs, and much more.

This series moves the vast majority of the builtin's functionality into several new compilation units, all grouped under a new repack.h API:

 * repack.c
 * repack-cruft.c
 * repack-filtered.c
 * repack-geometry.c
 * repack-midx.c
 * repack-promisor.c

Moving this code around was less than straightforward, hence the length of this series. Likewise, since the repacking code can be rather fragile, I chose to go slower here than I otherwise would have (e.g., almost all code movement changes are strict moves with no rewrites in between).

The structure of the series is roughly as follows:
 - The first 10 patches reduce the usage of the_repository, and
   the_hash_algo. They don't quite get us to being able to compile
   without USE_THE_REPOSITORY_VARIABLE for reasons described in the last
   commit of this section.
 - The next four commits introduces the new repack.o compilation unit,
   and moves a fairly isolated component of the repack builtin.
 - The next five commits move the existing_packs API from out of the
   builtin into the repack.h API.
 - The next five move the generated_pack API.
 - The next two move the repack_promisor_objects() API.
 - The next four move the pack_geometry API.
 - The next nine move various MIDX-related functionality out of the
   builtin.
 - The next nine move remaining functions that write packs out of the
   repack builtin.
 - The final commit cleans up the list of `#include`s within the
   builtin.

My hope is that this produces a series that is easier to review, despite its rather long length. Each individual commit tries to do as little as possible, and most will benefit from being reviewed with --color-moved. The result, in my opinion, is a 'builtin/repack.c' is much cleaner than it was when we started.

Nothing in this series is a strict requirement for implementing a repacking strategy based on incremental MIDX/bitmaps. But since the new functionality required for that is lengthy in and of itself, having a clean slate to start from seemed worthwhile to me.

Thanks in advance for your review!
Taylor Blau (49):
  builtin/repack.c: avoid "the_repository" in `cmd_repack()`
  builtin/repack.c: avoid "the_repository" in existing packs API
  builtin/repack.c: avoid "the_repository" when taking a ref snapshot
  builtin/repack.c: avoid "the_repository" when removing packs
  builtin/repack.c: avoid "the_repository" when repacking promisor
    objects
  builtin/repack.c: avoid "the_hash_algo" when deleting packs
  builtin/repack.c: avoid "the_hash_algo" in `write_oid()`
  builtin/repack: avoid "the_hash_algo" in `repack_promisor_objects()`
  builtin/repack.c: avoid "the_hash_algo" in `finish_pack_objects_cmd()`
  builtin/repack.c: avoid using `hash_to_hex()` in pack geometry
  repack: introduce new compilation unit
  builtin/repack.c: pass both pack_objects args to repack_config
  repack: move 'delta_base_offset' to 'struct pack_objects_args'
  repack: remove 'prepare_pack_objects' from the builtin
  builtin/repack.c: rename many 'struct existing_packs' functions
  repack: remove 'remove_redundant_pack' from the builtin
  builtin/repack.c: pass "packdir" when removing packs
  builtin/repack.c: avoid unnecessary numeric casts in existing_packs
  repack: remove 'existing_packs' API from the builtin
  builtin/repack.c: rename "struct generated_pack_data"
  builtin/repack.c: factor our "generated_pack_install"
  builtin/repack.c: pass "packtmp" to `generated_pack_populate()`
  builtin/repack.c: provide pack locations to `generated_pack_install()`
  repack: remove 'generated_pack' API from the builtin
  builtin/repack.c: pass "packtmp" to `repack_promisor_objects()`
  builtin/repack.c: remove "repack_promisor_objects()" from the builtin
  builtin/repack.c: rename various pack_geometry functions
  builtin/repack.c: pass 'pack_kept_objects' to `pack_geometry_init()`
  builtin/repack.c: pass 'packdir' to `pack_geometry_remove_redundant()`
  repack: remove pack_geometry API from the builtin
  builtin/repack.c: remove ref snapshotting from builtin
  builtin/repack.c: extract opts struct for
    'write_midx_included_packs()'
  builtin/repack.c: use a string_list for 'midx_pack_names'
  repack: keep track of MIDX pack names using existing_packs
  builtin/repack.c: reorder `remove_redundant_bitmaps()`
  builtin/repack.c: inline `remove_redundant_bitmaps()`
  builtin/repack.c: pass `repack_write_midx_opts` to
    `midx_included_packs`
  builtin/repack.c: inline packs within `write_midx_included_packs()`
  repack: 'write_midx_included_packs' API from the builtin
  builtin/repack.c: introduce `struct write_pack_opts`
  builtin/repack.c: use `write_pack_opts` within `write_cruft_pack()`
  repack: move `find_pack_prefix()` out of the builtin
  repack: extract `write_pack_opts_is_local()`
  builtin/repack.c: pass `write_pack_opts` to
    `finish_pack_objects_cmd()`
  repack: move `finish_pack_objects_cmd()` out of the builtin
  repack: move `pack_kept_objects` to `struct pack_objects_args`
  repack: move `write_filtered_pack()` out of the builtin
  repack: move `write_cruft_pack()` out of the builtin
  builtin/repack.c: clean up unused `#include`s
 Makefile          |    6 +
 builtin/repack.c  | 1348 +++------------------------------------------
 meson.build       |    6 +
 repack-cruft.c    |   99 ++++
 repack-filtered.c |   51 ++
 repack-geometry.c |  232 ++++++++
 repack-midx.c     |  372 +++++++++++++
 repack-promisor.c |  102 ++++
 repack.c          |  361 ++++++++++++
 repack.h          |  146 +++++
 10 files changed, 1461 insertions(+), 1262 deletions(-)
 create mode 100644 repack-cruft.c
 create mode 100644 repack-filtered.c
 create mode 100644 repack-geometry.c
 create mode 100644 repack-midx.c
 create mode 100644 repack-promisor.c
 create mode 100644 repack.c
 create mode 100644 repack.h
base-commit: dd52a29b78d80e425be660f3b443a42e0374a7d1
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 01/49] builtin/repack.c: avoid "the_repository" in `cmd_repack()`

Reduce builtin/repack.c's reliance on `the_repository` by using the currently-UNUSED "repo" parameter within cmd_repack().

The following commits will continue to reduce the usage of the_repository in other places within builtin/repack.c.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 31 ++++++++++++++++---------------
 1 file changed, 16 insertions(+), 15 deletions(-)
Show changes to builtin/repack.c +16 −15
diff --git a/builtin/repack.c b/builtin/repack.c
index e8730808c5..305782b2c9 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -1247,7 +1247,7 @@ static const char *find_pack_prefix(const char *packdir, const char *packtmp)
 int cmd_repack(int argc,
 	       const char **argv,
 	       const char *prefix,
-	       struct repository *repo UNUSED)
+	       struct repository *repo)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
@@ -1344,7 +1344,7 @@ int cmd_repack(int argc,
 
 	list_objects_filter_init(&po_args.filter_options);
 
-	repo_config(the_repository, repack_config, &cruft_po_args);
+	repo_config(repo, repack_config, &cruft_po_args);
 
 	argc = parse_options(argc, argv, prefix, builtin_repack_options,
 				git_repack_usage, 0);
@@ -1354,7 +1354,7 @@ int cmd_repack(int argc,
 	po_args.depth = xstrdup_or_null(opt_depth);
 	po_args.threads = xstrdup_or_null(opt_threads);
 
-	if (delete_redundant && the_repository->repository_format_precious_objects)
+	if (delete_redundant && repo->repository_format_precious_objects)
 		die(_("cannot delete packs in a precious-objects repo"));
 
 	die_for_incompatible_opt3(unpack_unreachable || (pack_everything & LOOSEN_UNREACHABLE), "-A",
@@ -1376,7 +1376,7 @@ int cmd_repack(int argc,
 		die(_(incremental_bitmap_conflict_error));
 
 	if (write_bitmaps && po_args.local &&
-	    odb_has_alternates(the_repository->objects)) {
+	    odb_has_alternates(repo->objects)) {
 		/*
 		 * When asked to do a local repack, but we have
 		 * packfiles that are inherited from an alternate, then
@@ -1391,7 +1391,8 @@ int cmd_repack(int argc,
 	if (write_midx && write_bitmaps) {
 		struct strbuf path = STRBUF_INIT;
 
-		strbuf_addf(&path, "%s/%s_XXXXXX", repo_get_object_directory(the_repository),
+		strbuf_addf(&path, "%s/%s_XXXXXX",
+			    repo_get_object_directory(repo),
 			    "bitmap-ref-tips");
 
 		refs_snapshot = xmks_tempfile(path.buf);
@@ -1400,7 +1401,7 @@ int cmd_repack(int argc,
 		strbuf_release(&path);
 	}
 
-	packdir = mkpathdup("%s/pack", repo_get_object_directory(the_repository));
+	packdir = mkpathdup("%s/pack", repo_get_object_directory(repo));
 	packtmp_name = xstrfmt(".tmp-%d-pack", (int)getpid());
 	packtmp = mkpathdup("%s/%s", packdir, packtmp_name);
 
@@ -1439,7 +1440,7 @@ int cmd_repack(int argc,
 		strvec_push(&cmd.args, "--reflog");
 		strvec_push(&cmd.args, "--indexed-objects");
 	}
-	if (repo_has_promisor_remote(the_repository))
+	if (repo_has_promisor_remote(repo))
 		strvec_push(&cmd.args, "--exclude-promisor-objects");
 	if (!write_midx) {
 		if (write_bitmaps > 0)
@@ -1535,7 +1536,7 @@ int cmd_repack(int argc,
 		 * midx_has_unknown_packs() will make the decision for
 		 * us.
 		 */
-		if (!get_multi_pack_index(the_repository->objects->sources))
+		if (!get_multi_pack_index(repo->objects->sources))
 			midx_must_contain_cruft = 1;
 	}
 
@@ -1618,9 +1619,9 @@ int cmd_repack(int argc,
 
 	string_list_sort(&names);
 
-	if (get_multi_pack_index(the_repository->objects->sources)) {
+	if (get_multi_pack_index(repo->objects->sources)) {
 		struct multi_pack_index *m =
-			get_multi_pack_index(the_repository->objects->sources);
+			get_multi_pack_index(repo->objects->sources);
 
 		ALLOC_ARRAY(midx_pack_names,
 			    m->num_packs + m->num_packs_in_base);
@@ -1631,7 +1632,7 @@ int cmd_repack(int argc,
 					xstrdup(m->pack_names[i]);
 	}
 
-	close_object_store(the_repository->objects);
+	close_object_store(repo->objects);
 
 	/*
 	 * Ok we have prepared all new packfiles.
@@ -1688,7 +1689,7 @@ int cmd_repack(int argc,
 			goto cleanup;
 	}
 
-	odb_reprepare(the_repository->objects);
+	odb_reprepare(repo->objects);
 
 	if (delete_redundant) {
 		int opts = 0;
@@ -1704,18 +1705,18 @@ int cmd_repack(int argc,
 		if (!keep_unreachable &&
 		    (!(pack_everything & LOOSEN_UNREACHABLE) ||
 		     unpack_unreachable) &&
-		    is_repository_shallow(the_repository))
+		    is_repository_shallow(repo))
 			prune_shallow(PRUNE_QUICK);
 	}
 
 	if (run_update_server_info)
-		update_server_info(the_repository, 0);
+		update_server_info(repo, 0);
 
 	if (git_env_bool(GIT_TEST_MULTI_PACK_INDEX, 0)) {
 		unsigned flags = 0;
 		if (git_env_bool(GIT_TEST_MULTI_PACK_INDEX_WRITE_INCREMENTAL, 0))
 			flags |= MIDX_WRITE_INCREMENTAL;
-		write_midx_file(the_repository->objects->sources,
+		write_midx_file(repo->objects->sources,
 				NULL, NULL, flags);
 	}
 
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 02/49] builtin/repack.c: avoid "the_repository" in existing packs API

There are a number of spots within builtin/repack.c which refer to "the_repository", and either make use of the "existing packs" API or otherwise have a 'struct existing_packs *' in scope.

Add a "repo" member to "struct existing_packs" and use that instead of "the_repository" in such locations.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
Show changes to builtin/repack.c +5 −3
diff --git a/builtin/repack.c b/builtin/repack.c
index 305782b2c9..7223553bed 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -126,6 +126,7 @@ static void pack_objects_args_release(struct pack_objects_args *args)
 }
 
 struct existing_packs {
+	struct repository *repo;
 	struct string_list kept_packs;
 	struct string_list non_kept_packs;
 	struct string_list cruft_packs;
@@ -265,7 +266,7 @@ static void existing_packs_release(struct existing_packs *existing)
 static void collect_pack_filenames(struct existing_packs *existing,
 				   const struct string_list *extra_keep)
 {
-	struct packfile_store *packs = the_repository->objects->packfiles;
+	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
 	struct strbuf buf = STRBUF_INIT;
 
@@ -498,7 +499,7 @@ static void init_pack_geometry(struct pack_geometry *geometry,
 			       struct existing_packs *existing,
 			       const struct pack_objects_args *args)
 {
-	struct packfile_store *packs = the_repository->objects->packfiles;
+	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
 	struct strbuf buf = STRBUF_INIT;
 
@@ -1139,7 +1140,7 @@ static int write_filtered_pack(const struct pack_objects_args *args,
 static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
 				      struct existing_packs *existing)
 {
-	struct packfile_store *packs = the_repository->objects->packfiles;
+	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
 	struct strbuf buf = STRBUF_INIT;
 	size_t i;
@@ -1405,6 +1406,7 @@ int cmd_repack(int argc,
 	packtmp_name = xstrfmt(".tmp-%d-pack", (int)getpid());
 	packtmp = mkpathdup("%s/%s", packdir, packtmp_name);
 
+	existing.repo = repo;
 	collect_pack_filenames(&existing, &keep_pack_list);
 
 	if (geometry.split_factor) {
-- 
2.51.0.243.g16eca91f2c0
Jeff King· Oct 10, 2025, 05:19 UTC · re: Taylor Blau · lore

Re: [PATCH 02/49] builtin/repack.c: avoid "the_repository" in existing packs API

On Sun, Sep 28, 2025 at 06:07:18PM -0400, Taylor Blau wrote:
Show 11 quoted lines
>  struct existing_packs {
> +	struct repository *repo;
>  	struct string_list kept_packs;
>  	struct string_list non_kept_packs;
>  	struct string_list cruft_packs;
> @@ -265,7 +266,7 @@ static void existing_packs_release(struct existing_packs *existing)
>  static void collect_pack_filenames(struct existing_packs *existing,
>  				   const struct string_list *extra_keep)
>  {
> -	struct packfile_store *packs = the_repository->objects->packfiles;
> +	struct packfile_store *packs = existing->repo->objects->packfiles;

I found it a little funny to pass around a repository struct as part of existing_packs, since they're not directly related. But I think it is mostly just a convenience to do so, and "existing_packs" is really storing an overall context.

It could probably be scoped down to pass around an object_database instead, or maybe even a packfile store, which somehow feels a little more "correct" to me as context. But I have trouble imagining that being helpful to any new code (why would it have an object_store but a not a repo object?). And I can easily imagine having the repo available being useful for future refactorings.

So I think what you've written here is probably a good path forward. I wanted to outline my thinking because I suspect the same will apply to other patches in this series (i.e., they don't always need a repo object, but it's simplest to pass them one).

-Peff
Taylor Blau· Oct 10, 2025, 22:48 UTC · re: Jeff King · lore

Re: [PATCH 02/49] builtin/repack.c: avoid "the_repository" in existing packs API

On Fri, Oct 10, 2025 at 01:19:55AM -0400, Jeff King wrote:
Show 18 quoted lines
> On Sun, Sep 28, 2025 at 06:07:18PM -0400, Taylor Blau wrote:
>
> >  struct existing_packs {
> > +	struct repository *repo;
> >  	struct string_list kept_packs;
> >  	struct string_list non_kept_packs;
> >  	struct string_list cruft_packs;
> > @@ -265,7 +266,7 @@ static void existing_packs_release(struct existing_packs *existing)
> >  static void collect_pack_filenames(struct existing_packs *existing,
> >  				   const struct string_list *extra_keep)
> >  {
> > -	struct packfile_store *packs = the_repository->objects->packfiles;
> > +	struct packfile_store *packs = existing->repo->objects->packfiles;
>
> I found it a little funny to pass around a repository struct as part of
> existing_packs, since they're not directly related. But I think it is
> mostly just a convenience to do so, and "existing_packs" is really
> storing an overall context.

Yeah, the patch as-is suggests that we could have gotten away with just adding a pointer to the packfile_store within existing_packs. I tried to do that locally, and it works up until "builtin/repack.c: rename many 'struct existing_packs' functions", where we start to want to use "repo", or "repo->hash_algo" :-<.

Show 11 quoted lines
> It could probably be scoped down to pass around an object_database
> instead, or maybe even a packfile store, which somehow feels a little
> more "correct" to me as context. But I have trouble imagining that being
> helpful to any new code (why would it have an object_store but a not a
> repo object?). And I can easily imagine having the repo available being
> useful for future refactorings.
>
> So I think what you've written here is probably a good path forward. I
> wanted to outline my thinking because I suspect the same will apply to
> other patches in this series (i.e., they don't always need a repo
> object, but it's simplest to pass them one).

Yeah, exactly. In general for this series I am trying to do as few changes to the actual functions as possible. There are some (like the renaming in the later patch I mentioned above), but only in service of trying to have a clean repack.h (at least from a naming standpoint).

There is lots of opportunity to clean up the repack code further. I agree it would be nice to have existing_packs only take a packfile_store, and pass around the repository in other places as needed. I'm happy to investigate that, but as a follow-up to this series, since I want to change as few things as possible here.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 03/49] builtin/repack.c: avoid "the_repository" when taking a ref snapshot

Avoid using "the_repository" in various MIDX-related ref snapshotting functions.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)
Show changes to builtin/repack.c +9 −7
diff --git a/builtin/repack.c b/builtin/repack.c
index 7223553bed..113f5fc67f 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -771,6 +771,7 @@ static int midx_has_unknown_packs(char **midx_pack_names,
 }
 
 struct midx_snapshot_ref_data {
+	struct repository *repo;
 	struct tempfile *f;
 	struct oidset seen;
 	int preferred;
@@ -784,13 +785,13 @@ static int midx_snapshot_ref_one(const char *refname UNUSED,
 	struct midx_snapshot_ref_data *data = _data;
 	struct object_id peeled;
 
-	if (!peel_iterated_oid(the_repository, oid, &peeled))
+	if (!peel_iterated_oid(data->repo, oid, &peeled))
 		oid = &peeled;
 
 	if (oidset_insert(&data->seen, oid))
 		return 0; /* already seen */
 
-	if (odb_read_object_info(the_repository->objects, oid, NULL) != OBJ_COMMIT)
+	if (odb_read_object_info(data->repo->objects, oid, NULL) != OBJ_COMMIT)
 		return 0;
 
 	fprintf(data->f->fp, "%s%s\n", data->preferred ? "+" : "",
@@ -799,11 +800,12 @@ static int midx_snapshot_ref_one(const char *refname UNUSED,
 	return 0;
 }
 
-static void midx_snapshot_refs(struct tempfile *f)
+static void midx_snapshot_refs(struct repository *repo, struct tempfile *f)
 {
 	struct midx_snapshot_ref_data data;
-	const struct string_list *preferred = bitmap_preferred_tips(the_repository);
+	const struct string_list *preferred = bitmap_preferred_tips(repo);
 
+	data.repo = repo;
 	data.f = f;
 	data.preferred = 0;
 	oidset_init(&data.seen, 0);
@@ -817,13 +819,13 @@ static void midx_snapshot_refs(struct tempfile *f)
 
 		data.preferred = 1;
 		for_each_string_list_item(item, preferred)
-			refs_for_each_ref_in(get_main_ref_store(the_repository),
+			refs_for_each_ref_in(get_main_ref_store(repo),
 					     item->string,
 					     midx_snapshot_ref_one, &data);
 		data.preferred = 0;
 	}
 
-	refs_for_each_ref(get_main_ref_store(the_repository),
+	refs_for_each_ref(get_main_ref_store(repo),
 			  midx_snapshot_ref_one, &data);
 
 	if (close_tempfile_gently(f)) {
@@ -1397,7 +1399,7 @@ int cmd_repack(int argc,
 			    "bitmap-ref-tips");
 
 		refs_snapshot = xmks_tempfile(path.buf);
-		midx_snapshot_refs(refs_snapshot);
+		midx_snapshot_refs(repo, refs_snapshot);
 
 		strbuf_release(&path);
 	}
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 04/49] builtin/repack.c: avoid "the_repository" when removing packs

The 'remove_redundant_pack()' function uses "the_repository" to obtain, and optionally remove, the repository's MIDX. Instead of relying on "the_repository", pass around a "struct repository *" parameter through its callers, and use that instead.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)
Show changes to builtin/repack.c +10 −8
diff --git a/builtin/repack.c b/builtin/repack.c
index 113f5fc67f..93802531e1 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -221,33 +221,35 @@ static void mark_packs_for_deletion(struct existing_packs *existing,
 	mark_packs_for_deletion_1(names, &existing->cruft_packs);
 }
 
-static void remove_redundant_pack(const char *dir_name, const char *base_name)
+static void remove_redundant_pack(struct repository *repo,
+				  const char *dir_name, const char *base_name)
 {
 	struct strbuf buf = STRBUF_INIT;
-	struct odb_source *source = the_repository->objects->sources;
+	struct odb_source *source = repo->objects->sources;
 	struct multi_pack_index *m = get_multi_pack_index(source);
 	strbuf_addf(&buf, "%s.pack", base_name);
 	if (m && source->local && midx_contains_pack(m, buf.buf))
-		clear_midx_file(the_repository);
+		clear_midx_file(repo);
 	strbuf_insertf(&buf, 0, "%s/", dir_name);
 	unlink_pack_path(buf.buf, 1);
 	strbuf_release(&buf);
 }
 
-static void remove_redundant_packs_1(struct string_list *packs)
+static void remove_redundant_packs_1(struct repository *repo,
+				     struct string_list *packs)
 {
 	struct string_list_item *item;
 	for_each_string_list_item(item, packs) {
 		if (!pack_is_marked_for_deletion(item))
 			continue;
-		remove_redundant_pack(packdir, item->string);
+		remove_redundant_pack(repo, packdir, item->string);
 	}
 }
 
 static void remove_redundant_existing_packs(struct existing_packs *existing)
 {
-	remove_redundant_packs_1(&existing->non_kept_packs);
-	remove_redundant_packs_1(&existing->cruft_packs);
+	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs);
+	remove_redundant_packs_1(existing->repo, &existing->cruft_packs);
 }
 
 static void existing_packs_release(struct existing_packs *existing)
@@ -685,7 +687,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
 		    (string_list_has_string(&existing->kept_packs, buf.buf)))
 			continue;
 
-		remove_redundant_pack(packdir, buf.buf);
+		remove_redundant_pack(existing->repo, packdir, buf.buf);
 	}
 
 	strbuf_release(&buf);
-- 
2.51.0.243.g16eca91f2c0
Jeff King· Oct 10, 2025, 05:22 UTC · re: Taylor Blau · lore

Re: [PATCH 04/49] builtin/repack.c: avoid "the_repository" when removing packs

On Sun, Sep 28, 2025 at 06:07:26PM -0400, Taylor Blau wrote:
Show 16 quoted lines
> -static void remove_redundant_pack(const char *dir_name, const char *base_name)
> +static void remove_redundant_pack(struct repository *repo,
> +				  const char *dir_name, const char *base_name)
>  {
>  	struct strbuf buf = STRBUF_INIT;
> -	struct odb_source *source = the_repository->objects->sources;
> +	struct odb_source *source = repo->objects->sources;
>  	struct multi_pack_index *m = get_multi_pack_index(source);
>  	strbuf_addf(&buf, "%s.pack", base_name);
>  	if (m && source->local && midx_contains_pack(m, buf.buf))
> -		clear_midx_file(the_repository);
> +		clear_midx_file(repo);
>  	strbuf_insertf(&buf, 0, "%s/", dir_name);
>  	unlink_pack_path(buf.buf, 1);
>  	strbuf_release(&buf);
>  }

Ah, yeah, this is a good example of what I was talking about in the other part of the thread. Probably this function could just take the object_source pointer to find the midx. But then it wouldn't be able to use clear_midx_file(). That function arguably should itself take an object_source pointer and not a repo pointer, but it feels like minimal gain to go around trying to tighten these (and certainly something we could easily do later, even programatically, if we cared).

-Peff
Taylor Blau· Oct 10, 2025, 22:49 UTC · re: Jeff King · lore

Re: [PATCH 04/49] builtin/repack.c: avoid "the_repository" when removing packs

On Fri, Oct 10, 2025 at 01:22:56AM -0400, Jeff King wrote:
Show 26 quoted lines
> On Sun, Sep 28, 2025 at 06:07:26PM -0400, Taylor Blau wrote:
>
> > -static void remove_redundant_pack(const char *dir_name, const char *base_name)
> > +static void remove_redundant_pack(struct repository *repo,
> > +				  const char *dir_name, const char *base_name)
> >  {
> >  	struct strbuf buf = STRBUF_INIT;
> > -	struct odb_source *source = the_repository->objects->sources;
> > +	struct odb_source *source = repo->objects->sources;
> >  	struct multi_pack_index *m = get_multi_pack_index(source);
> >  	strbuf_addf(&buf, "%s.pack", base_name);
> >  	if (m && source->local && midx_contains_pack(m, buf.buf))
> > -		clear_midx_file(the_repository);
> > +		clear_midx_file(repo);
> >  	strbuf_insertf(&buf, 0, "%s/", dir_name);
> >  	unlink_pack_path(buf.buf, 1);
> >  	strbuf_release(&buf);
> >  }
>
> Ah, yeah, this is a good example of what I was talking about in the
> other part of the thread. Probably this function could just take the
> object_source pointer to find the midx. But then it wouldn't be able to
> use clear_midx_file(). That function arguably should itself take an
> object_source pointer and not a repo pointer, but it feels like minimal
> gain to go around trying to tighten these (and certainly something we
> could easily do later, even programatically, if we cared).

Yup, I agree. The changes to `clear_midx_file()` seem like a good candidate for #leftoverbits to me.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 05/49] builtin/repack.c: avoid "the_repository" when repacking promisor objects

Pass a "struct repository" pointer to the 'repack_promisor_objects()' function to avoid using "the_repository".

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
Show changes to builtin/repack.c +4 −3
diff --git a/builtin/repack.c b/builtin/repack.c
index 93802531e1..4f08b57ddb 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -407,7 +407,8 @@ static int has_pack_ext(const struct generated_pack_data *data,
 	BUG("unknown pack extension: '%s'", ext);
 }
 
-static void repack_promisor_objects(const struct pack_objects_args *args,
+static void repack_promisor_objects(struct repository *repo,
+				    const struct pack_objects_args *args,
 				    struct string_list *names)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
@@ -424,7 +425,7 @@ static void repack_promisor_objects(const struct pack_objects_args *args,
 	 * {type -> existing pack order} ordering when computing deltas instead
 	 * of a {type -> size} ordering, which may produce better deltas.
 	 */
-	for_each_packed_object(the_repository, write_oid, &cmd,
+	for_each_packed_object(repo, write_oid, &cmd,
 			       FOR_EACH_OBJECT_PROMISOR_ONLY);
 
 	if (cmd.in == -1) {
@@ -1458,7 +1459,7 @@ int cmd_repack(int argc,
 		strvec_push(&cmd.args, "--delta-islands");
 
 	if (pack_everything & ALL_INTO_ONE) {
-		repack_promisor_objects(&po_args, &names);
+		repack_promisor_objects(repo, &po_args, &names);
 
 		if (has_existing_non_kept_packs(&existing) &&
 		    delete_redundant &&
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 06/49] builtin/repack.c: avoid "the_hash_algo" when deleting packs

The "mark_packs_for_deletion_1" function uses "the_hash_algo->hexsz" to isolate a pack's checksum before deleting it to avoid deleting a newly written pack having the same checksum (that is, some generated pack wound up identical to an existing pack).

Avoid this by passing down a "struct git_hash_algo" pointer, and refer to the hash algorithm through it instead.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
Show changes to builtin/repack.c +6 −4
diff --git a/builtin/repack.c b/builtin/repack.c
index 4f08b57ddb..094f5a0cc2 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -168,11 +168,12 @@ static int pack_is_retained(struct string_list_item *item)
 	return (uintptr_t)item->util & RETAIN_PACK;
 }
 
-static void mark_packs_for_deletion_1(struct string_list *names,
+static void mark_packs_for_deletion_1(const struct git_hash_algo *algop,
+				      struct string_list *names,
 				      struct string_list *list)
 {
 	struct string_list_item *item;
-	const int hexsz = the_hash_algo->hexsz;
+	const int hexsz = algop->hexsz;
 
 	for_each_string_list_item(item, list) {
 		char *sha1;
@@ -217,8 +218,9 @@ static void mark_packs_for_deletion(struct existing_packs *existing,
 				    struct string_list *names)
 
 {
-	mark_packs_for_deletion_1(names, &existing->non_kept_packs);
-	mark_packs_for_deletion_1(names, &existing->cruft_packs);
+	const struct git_hash_algo *algop = existing->repo->hash_algo;
+	mark_packs_for_deletion_1(algop, names, &existing->non_kept_packs);
+	mark_packs_for_deletion_1(algop, names, &existing->cruft_packs);
 }
 
 static void remove_redundant_pack(struct repository *repo,
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 07/49] builtin/repack.c: avoid "the_hash_algo" in `write_oid()`

In a similar spirit as the previous commit, avoid referring directly to "the_hash_algo" within builtin/repack.c::write_oid().

Unlike the previous commit, we are within a callback function, so must introduce a new struct to pass additional data through its "data" pointer.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)
Show changes to builtin/repack.c +12 −3
diff --git a/builtin/repack.c b/builtin/repack.c
index 094f5a0cc2..7d62959dc2 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -339,6 +339,11 @@ static void prepare_pack_objects(struct child_process *cmd,
 	cmd->out = -1;
 }
 
+struct write_oid_context {
+	struct child_process *cmd;
+	const struct git_hash_algo *algop;
+};
+
 /*
  * Write oid to the given struct child_process's stdin, starting it first if
  * necessary.
@@ -347,14 +352,15 @@ static int write_oid(const struct object_id *oid,
 		     struct packed_git *pack UNUSED,
 		     uint32_t pos UNUSED, void *data)
 {
-	struct child_process *cmd = data;
+	struct write_oid_context *ctx = data;
+	struct child_process *cmd = ctx->cmd;
 
 	if (cmd->in == -1) {
 		if (start_command(cmd))
 			die(_("could not start pack-objects to repack promisor objects"));
 	}
 
-	if (write_in_full(cmd->in, oid_to_hex(oid), the_hash_algo->hexsz) < 0 ||
+	if (write_in_full(cmd->in, oid_to_hex(oid), ctx->algop->hexsz) < 0 ||
 	    write_in_full(cmd->in, "\n", 1) < 0)
 		die(_("failed to feed promisor objects to pack-objects"));
 	return 0;
@@ -413,6 +419,7 @@ static void repack_promisor_objects(struct repository *repo,
 				    const struct pack_objects_args *args,
 				    struct string_list *names)
 {
+	struct write_oid_context ctx;
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	FILE *out;
 	struct strbuf line = STRBUF_INIT;
@@ -427,7 +434,9 @@ static void repack_promisor_objects(struct repository *repo,
 	 * {type -> existing pack order} ordering when computing deltas instead
 	 * of a {type -> size} ordering, which may produce better deltas.
 	 */
-	for_each_packed_object(repo, write_oid, &cmd,
+	ctx.cmd = &cmd;
+	ctx.algop = repo->hash_algo;
+	for_each_packed_object(repo, write_oid, &ctx,
 			       FOR_EACH_OBJECT_PROMISOR_ONLY);
 
 	if (cmd.in == -1) {
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Sep 29, 2025, 23:21 UTC · re: Taylor Blau · lore

Re: [PATCH 07/49] builtin/repack.c: avoid "the_hash_algo" in `write_oid()`

On Sun, Sep 28, 2025 at 06:07:37PM -0400, Taylor Blau wrote:
Show 21 quoted lines
> diff --git a/builtin/repack.c b/builtin/repack.c
> index 094f5a0cc2..7d62959dc2 100644
> --- a/builtin/repack.c
> +++ b/builtin/repack.c
> @@ -413,6 +419,7 @@ static void repack_promisor_objects(struct repository *repo,
>  				    const struct pack_objects_args *args,
>  				    struct string_list *names)
>  {
> +	struct write_oid_context ctx;
>  	struct child_process cmd = CHILD_PROCESS_INIT;
>  	FILE *out;
>  	struct strbuf line = STRBUF_INIT;
> @@ -427,7 +434,9 @@ static void repack_promisor_objects(struct repository *repo,
>  	 * {type -> existing pack order} ordering when computing deltas instead
>  	 * of a {type -> size} ordering, which may produce better deltas.
>  	 */
> -	for_each_packed_object(repo, write_oid, &cmd,
> +	ctx.cmd = &cmd;
> +	ctx.algop = repo->hash_algo;
> +	for_each_packed_object(repo, write_oid, &ctx,
>  			       FOR_EACH_OBJECT_PROMISOR_ONLY);

As this one is essentially new code, and we have all of the info available right at the start of the problem. Do we maybe want to use designated initializers?

Patrick
Taylor Blau· Oct 7, 2025, 20:17 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 07/49] builtin/repack.c: avoid "the_hash_algo" in `write_oid()`

On Tue, Sep 30, 2025 at 01:21:10AM +0200, Patrick Steinhardt wrote:
Show 26 quoted lines
> On Sun, Sep 28, 2025 at 06:07:37PM -0400, Taylor Blau wrote:
> > diff --git a/builtin/repack.c b/builtin/repack.c
> > index 094f5a0cc2..7d62959dc2 100644
> > --- a/builtin/repack.c
> > +++ b/builtin/repack.c
> > @@ -413,6 +419,7 @@ static void repack_promisor_objects(struct repository *repo,
> >  				    const struct pack_objects_args *args,
> >  				    struct string_list *names)
> >  {
> > +	struct write_oid_context ctx;
> >  	struct child_process cmd = CHILD_PROCESS_INIT;
> >  	FILE *out;
> >  	struct strbuf line = STRBUF_INIT;
> > @@ -427,7 +434,9 @@ static void repack_promisor_objects(struct repository *repo,
> >  	 * {type -> existing pack order} ordering when computing deltas instead
> >  	 * of a {type -> size} ordering, which may produce better deltas.
> >  	 */
> > -	for_each_packed_object(repo, write_oid, &cmd,
> > +	ctx.cmd = &cmd;
> > +	ctx.algop = repo->hash_algo;
> > +	for_each_packed_object(repo, write_oid, &ctx,
> >  			       FOR_EACH_OBJECT_PROMISOR_ONLY);
>
> As this one is essentially new code, and we have all of the info
> available right at the start of the problem. Do we maybe want to use
> designated initializers?

I considered it, but ultimately decided against it here because I find it clearer to initialize "ctx" right before it is used to avoid ambiguity.

Alternatively, you could accomplish this with a compound literal, but that's (a) still a test balloon and (b) more verbose than necessary.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 08/49] builtin/repack: avoid "the_hash_algo" in `repack_promisor_objects()`

In a similar spirit as the previous commits, avoid referring directly to "the_hash_algo" within builtin/repack.c::repack_promisor_objects().

Since there is already a repository pointer in scope, use its hash_algo value instead.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Show changes to builtin/repack.c +1 −1
diff --git a/builtin/repack.c b/builtin/repack.c
index 7d62959dc2..a7e94ed03c 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -452,7 +452,7 @@ static void repack_promisor_objects(struct repository *repo,
 		struct string_list_item *item;
 		char *promisor_name;
 
-		if (line.len != the_hash_algo->hexsz)
+		if (line.len != repo->hash_algo->hexsz)
 			die(_("repack: Expecting full hex object ID lines only from pack-objects."));
 		item = string_list_append(names, line.buf);
 
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 09/49] builtin/repack.c: avoid "the_hash_algo" in `finish_pack_objects_cmd()`

In a similar spirit as previous commits, avoid referring directly to "the_hash_algo" in builtin/repack.c::finish_pack_objects_cmd() and instead accept one as a parameter to the function.

Since this function has a number of callers throughout the builtin, the diff is a little noisier than previous commits. However, each hunk is limited to passing the hash_algo parameter from a repository pointer that is already in scope.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)
Show changes to builtin/repack.c +8 −5
diff --git a/builtin/repack.c b/builtin/repack.c
index a7e94ed03c..a043704aa8 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -1073,7 +1073,8 @@ static void remove_redundant_bitmaps(struct string_list *include,
 	strbuf_release(&path);
 }
 
-static int finish_pack_objects_cmd(struct child_process *cmd,
+static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
+				   struct child_process *cmd,
 				   struct string_list *names,
 				   int local)
 {
@@ -1084,7 +1085,7 @@ static int finish_pack_objects_cmd(struct child_process *cmd,
 	while (strbuf_getline_lf(&line, out) != EOF) {
 		struct string_list_item *item;
 
-		if (line.len != the_hash_algo->hexsz)
+		if (line.len != algop->hexsz)
 			die(_("repack: Expecting full hex object ID lines only "
 			      "from pack-objects."));
 		/*
@@ -1150,7 +1151,8 @@ static int write_filtered_pack(const struct pack_objects_args *args,
 		fprintf(in, "%s%s.pack\n", caret, item->string);
 	fclose(in);
 
-	return finish_pack_objects_cmd(&cmd, names, local);
+	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
+				       local);
 }
 
 static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
@@ -1247,7 +1249,8 @@ static int write_cruft_pack(const struct pack_objects_args *args,
 		fprintf(in, "%s.pack\n", item->string);
 	fclose(in);
 
-	return finish_pack_objects_cmd(&cmd, names, local);
+	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
+				       local);
 }
 
 static const char *find_pack_prefix(const char *packdir, const char *packtmp)
@@ -1534,7 +1537,7 @@ int cmd_repack(int argc,
 		fclose(in);
 	}
 
-	ret = finish_pack_objects_cmd(&cmd, &names, 1);
+	ret = finish_pack_objects_cmd(repo->hash_algo, &cmd, &names, 1);
 	if (ret)
 		goto cleanup;
 
-- 
2.51.0.243.g16eca91f2c0
Jeff King· Oct 10, 2025, 05:31 UTC · re: Taylor Blau · lore

Re: [PATCH 09/49] builtin/repack.c: avoid "the_hash_algo" in `finish_pack_objects_cmd()`

On Sun, Sep 28, 2025 at 06:07:45PM -0400, Taylor Blau wrote:
Show 8 quoted lines
> @@ -1150,7 +1151,8 @@ static int write_filtered_pack(const struct pack_objects_args *args,
>  		fprintf(in, "%s%s.pack\n", caret, item->string);
>  	fclose(in);
>  
> -	return finish_pack_objects_cmd(&cmd, names, local);
> +	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
> +				       local);
>  }

OK, my last musing on which types to pass around, I promise. ;) This is a great counter-example to my initial comment on patch 2: if we had put an object_source into the "existing" struct back then, we wouldn't be able to pull the hash_algo out now.

(Well, sort of; object_database does have a pointer back to its containing repo, which itself feels kind of weird. I'd just as soon not rely on that, here, though).

> -static int finish_pack_objects_cmd(struct child_process *cmd,
> +static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
> +				   struct child_process *cmd,

I am mildly surprised at some of these spots that _only_ need a hash_algo and nothing else, but the proof of the pudding is in the compiling. As they say. Well, as I say.

-Peff
Taylor Blau· Oct 10, 2025, 22:51 UTC · re: Jeff King · lore

Re: [PATCH 09/49] builtin/repack.c: avoid "the_hash_algo" in `finish_pack_objects_cmd()`

On Fri, Oct 10, 2025 at 01:31:15AM -0400, Jeff King wrote:
Show 19 quoted lines
> On Sun, Sep 28, 2025 at 06:07:45PM -0400, Taylor Blau wrote:
>
> > @@ -1150,7 +1151,8 @@ static int write_filtered_pack(const struct pack_objects_args *args,
> >  		fprintf(in, "%s%s.pack\n", caret, item->string);
> >  	fclose(in);
> >
> > -	return finish_pack_objects_cmd(&cmd, names, local);
> > +	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
> > +				       local);
> >  }
>
> OK, my last musing on which types to pass around, I promise. ;) This is
> a great counter-example to my initial comment on patch 2: if we had put
> an object_source into the "existing" struct back then, we wouldn't be
> able to pull the hash_algo out now.
>
> (Well, sort of; object_database does have a pointer back to its
> containing repo, which itself feels kind of weird. I'd just as soon not
> rely on that, here, though).

Yeah, that feels like a tinier example of what we're talking about here, just within the object_database struct itself ;-). I agree that it does not make sense to rely on that, since I think the existence of that pointer is equally something that we could consider changing.

Show 7 quoted lines
> > -static int finish_pack_objects_cmd(struct child_process *cmd,
> > +static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
> > +				   struct child_process *cmd,
>
> I am mildly surprised at some of these spots that _only_ need a
> hash_algo and nothing else, but the proof of the pudding is in the
> compiling. As they say. Well, as I say.
I would like that on a t-shirt, please.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 10/49] builtin/repack.c: avoid using `hash_to_hex()` in pack geometry

In previous commits, we started passing either repository or git_hash_algo pointers around to various spots within builtin/repack.c to reduce our dependency on the_repository in the hope of undef'ing USE_THE_REPOSITORY_VARIABLE.

This commit takes us as far as we can (easily) go in that direction by removing the only use of a convenience function that only exists when USE_THE_REPOSITORY_VARIABLE is defined.

Unfortunately, the only other such function is "is_bare_repository()", which is less than straightforward to convert into, say, "repo_is_bare()", the latter of the two accepting a repository pointer.

Punt on that for now, and declare this commit as the stopping point for our efforts in the direction of undef'ing USE_THE_REPOSITORY_VARIABLE.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Show changes to builtin/repack.c +3 −1
diff --git a/builtin/repack.c b/builtin/repack.c
index a043704aa8..0d35f15b4b 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -683,12 +683,14 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
 					    struct string_list *names,
 					    struct existing_packs *existing)
 {
+	const struct git_hash_algo *algop = existing->repo->hash_algo;
 	struct strbuf buf = STRBUF_INIT;
 	uint32_t i;
 
 	for (i = 0; i < geometry->split; i++) {
 		struct packed_git *p = geometry->pack[i];
-		if (string_list_has_string(names, hash_to_hex(p->hash)))
+		if (string_list_has_string(names, hash_to_hex_algop(p->hash,
+								    algop)))
 			continue;
 
 		strbuf_reset(&buf);
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Sep 29, 2025, 23:21 UTC · re: Taylor Blau · lore

Re: [PATCH 10/49] builtin/repack.c: avoid using `hash_to_hex()` in pack geometry

On Sun, Sep 28, 2025 at 06:07:49PM -0400, Taylor Blau wrote:
Show 12 quoted lines
> In previous commits, we started passing either repository or
> git_hash_algo pointers around to various spots within builtin/repack.c
> to reduce our dependency on the_repository in the hope of undef'ing
> USE_THE_REPOSITORY_VARIABLE.
> 
> This commit takes us as far as we can (easily) go in that direction by
> removing the only use of a convenience function that only exists when
> USE_THE_REPOSITORY_VARIABLE is defined.
> 
> Unfortunately, the only other such function is "is_bare_repository()",
> which is less than straightforward to convert into, say,
> "repo_is_bare()", the latter of the two accepting a repository pointer.

Yeah, this function has been annoying me for quite a while already. It's definitely non-trivial to convert, so punting it into the future makes sense to me.

Patrick
Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 11/49] repack: introduce new compilation unit

Over the years, builtin/repack.c has turned into a grab-bag of functionality powering the 'git repack' builtin. Among its many capabilities, it:

 - can build and spawn 'git pack-objects' commands, which in turn
   generate new packs
 - has infrastructure to manage the set of existing packs in a
   repository
 - has infrastructure to split a sequence of packs into a geometric
   progression based on object size
 - can manage both generating and combining cruft packs together
 - can write new MIDXs
to name a few.

As a result, this builtin has accumulated a lot of code, making adding new functionality difficult. In the future, 'repack' will learn how to manage a chain of incremental MIDXs, adding yet more functionality into the builtin.

As a prerequisite step, let's first move some of the functionality in the builtin into its own repack.[ch].

This will be done over the course of many steps, since there are many individual components, some of which will end up in other, yet-to-exist compilation units of their own. Some of the code movement here is also non-trivial, so performing it in individual steps will make it easier to verify.

Let's start by migrating 'struct pack_objects_args' (and the related corresponding pack_objects_args_release() function) into repack.h, and teach both the Makefile and Meson how to build the new compilation unit.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile         |  1 +
 builtin/repack.c | 25 +------------------------
 meson.build      |  1 +
 repack.c         | 11 +++++++++++
 repack.h         | 23 +++++++++++++++++++++++
 5 files changed, 37 insertions(+), 24 deletions(-)
 create mode 100644 repack.c
 create mode 100644 repack.h
Show changes to 5 files +37 −24

Makefile, builtin/repack.c, meson.build, repack.c, repack.h

diff --git a/Makefile b/Makefile
index 555b7f4dc3..b220f0f7d8 100644
--- a/Makefile
+++ b/Makefile
@@ -1136,6 +1136,7 @@ LIB_OBJS += refs/packed-backend.o
 LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
+LIB_OBJS += repack.o
 LIB_OBJS += replace-object.o
 LIB_OBJS += repo-settings.o
 LIB_OBJS += repository.o
diff --git a/builtin/repack.c b/builtin/repack.c
index 0d35f15b4b..6dfcb3327e 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -19,6 +19,7 @@
 #include "prune-packed.h"
 #include "odb.h"
 #include "promisor-remote.h"
+#include "repack.h"
 #include "shallow.h"
 #include "pack.h"
 #include "pack-bitmap.h"
@@ -53,21 +54,6 @@ static const char incremental_bitmap_conflict_error[] = N_(
 "--no-write-bitmap-index or disable the pack.writeBitmaps configuration."
 );
 
-struct pack_objects_args {
-	char *window;
-	char *window_memory;
-	char *depth;
-	char *threads;
-	unsigned long max_pack_size;
-	int no_reuse_delta;
-	int no_reuse_object;
-	int quiet;
-	int local;
-	int name_hash_version;
-	int path_walk;
-	struct list_objects_filter_options filter_options;
-};
-
 static int repack_config(const char *var, const char *value,
 			 const struct config_context *ctx, void *cb)
 {
@@ -116,15 +102,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-static void pack_objects_args_release(struct pack_objects_args *args)
-{
-	free(args->window);
-	free(args->window_memory);
-	free(args->depth);
-	free(args->threads);
-	list_objects_filter_release(&args->filter_options);
-}
-
 struct existing_packs {
 	struct repository *repo;
 	struct string_list kept_packs;
diff --git a/meson.build b/meson.build
index e8ec0eca16..8ab287a2ed 100644
--- a/meson.build
+++ b/meson.build
@@ -462,6 +462,7 @@ libgit_sources = [
   'reftable/tree.c',
   'reftable/writer.c',
   'remote.c',
+  'repack.c',
   'replace-object.c',
   'repo-settings.c',
   'repository.c',
diff --git a/repack.c b/repack.c
new file mode 100644
index 0000000000..a1f5b796fb
--- /dev/null
+++ b/repack.c
@@ -0,0 +1,11 @@
+#include "git-compat-util.h"
+#include "repack.h"
+
+void pack_objects_args_release(struct pack_objects_args *args)
+{
+	free(args->window);
+	free(args->window_memory);
+	free(args->depth);
+	free(args->threads);
+	list_objects_filter_release(&args->filter_options);
+}
diff --git a/repack.h b/repack.h
new file mode 100644
index 0000000000..421d439d5a
--- /dev/null
+++ b/repack.h
@@ -0,0 +1,23 @@
+#ifndef REPACK_H
+#define REPACK_H
+
+#include "list-objects-filter-options.h"
+
+struct pack_objects_args {
+	char *window;
+	char *window_memory;
+	char *depth;
+	char *threads;
+	unsigned long max_pack_size;
+	int no_reuse_delta;
+	int no_reuse_object;
+	int quiet;
+	int local;
+	int name_hash_version;
+	int path_walk;
+	struct list_objects_filter_options filter_options;
+};
+
+void pack_objects_args_release(struct pack_objects_args *args);
+
+#endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Jeff King· Oct 10, 2025, 05:48 UTC · re: Taylor Blau · lore

Re: [PATCH 11/49] repack: introduce new compilation unit

On Sun, Sep 28, 2025 at 06:07:53PM -0400, Taylor Blau wrote:
> Let's start by migrating 'struct pack_objects_args' (and the related
> corresponding pack_objects_args_release() function) into repack.h, and
> teach both the Makefile and Meson how to build the new compilation unit.

This name is a little generic for something that now has external scope, since it is specific to repack's view of what it might want to pass to pack_objects (and not something that builtin/pack-objects.c cares about itself).

I don't have a great alternative name, though. I guess it could be repack_pack_objects_args or something, though that is getting to be a mouthful.

I dunno. Maybe it is OK as-is. If I stretch I can imagine a world where pack_objects_args is for the lib-ified version of cmd_pack_objects() to use (like we have send_pack_args, etc). But TBH I find it unlikely that we'd go that way anytime soon (and if we did, we'd maybe even be able to refactor repack to more directly use the struct offered by pack-objects).

-Peff
Taylor Blau· Oct 10, 2025, 22:52 UTC · re: Jeff King · lore

Re: [PATCH 11/49] repack: introduce new compilation unit

On Fri, Oct 10, 2025 at 01:48:04AM -0400, Jeff King wrote:
Show 14 quoted lines
> On Sun, Sep 28, 2025 at 06:07:53PM -0400, Taylor Blau wrote:
>
> > Let's start by migrating 'struct pack_objects_args' (and the related
> > corresponding pack_objects_args_release() function) into repack.h, and
> > teach both the Makefile and Meson how to build the new compilation unit.
>
> This name is a little generic for something that now has external scope,
> since it is specific to repack's view of what it might want to pass to
> pack_objects (and not something that builtin/pack-objects.c cares about
> itself).
>
> I don't have a great alternative name, though. I guess it could be
> repack_pack_objects_args or something, though that is getting to be a
> mouthful.

Yeah, I had considered `repack_pack_objects_args` initially for exactly that reason, but discarded it because I thought it was too wordy.

Show 6 quoted lines
> I dunno. Maybe it is OK as-is. If I stretch I can imagine a world where
> pack_objects_args is for the lib-ified version of cmd_pack_objects() to
> use (like we have send_pack_args, etc). But TBH I find it unlikely that
> we'd go that way anytime soon (and if we did, we'd maybe even be able to
> refactor repack to more directly use the struct offered by
> pack-objects).

I had a similar concern when writing this patch. I think that that world is far enough away that I am comfortable crossing that bridge if/when we get there ;-).

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:07 UTC · re: Taylor Blau · lore

[PATCH 12/49] builtin/repack.c: pass both pack_objects args to repack_config

A subsequent commit will remove 'delta_base_offset' as a static variable within builtin/repack.c, and reintroduce it as a member of the 'struct pack_objects_args'.

As a result, the repack_config callback will need to have both the cruft- and non-cruft 'struct pack_objects_args's in scope. Introduce a new 'struct repack_config_ctx' to allow the callee to provide both pointers to the callback.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +13 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index 6dfcb3327e..af6de8d77a 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -54,10 +54,16 @@ static const char incremental_bitmap_conflict_error[] = N_(
 "--no-write-bitmap-index or disable the pack.writeBitmaps configuration."
 );
 
+struct repack_config_ctx {
+	struct pack_objects_args *po_args;
+	struct pack_objects_args *cruft_po_args;
+};
+
 static int repack_config(const char *var, const char *value,
 			 const struct config_context *ctx, void *cb)
 {
-	struct pack_objects_args *cruft_po_args = cb;
+	struct repack_config_ctx *repack_ctx = cb;
+	struct pack_objects_args *cruft_po_args = repack_ctx->cruft_po_args;
 	if (!strcmp(var, "repack.usedeltabaseoffset")) {
 		delta_base_offset = git_config_bool(var, value);
 		return 0;
@@ -1260,6 +1266,7 @@ int cmd_repack(int argc,
 	size_t midx_pack_names_nr = 0;
 
 	/* variables to be filled by option parsing */
+	struct repack_config_ctx config_ctx;
 	int delete_redundant = 0;
 	const char *unpack_unreachable = NULL;
 	int keep_unreachable = 0;
@@ -1343,7 +1350,11 @@ int cmd_repack(int argc,
 
 	list_objects_filter_init(&po_args.filter_options);
 
-	repo_config(repo, repack_config, &cruft_po_args);
+	memset(&config_ctx, 0, sizeof(config_ctx));
+	config_ctx.po_args = &po_args;
+	config_ctx.cruft_po_args = &cruft_po_args;
+
+	repo_config(repo, repack_config, &config_ctx);
 
 	argc = parse_options(argc, argv, prefix, builtin_repack_options,
 				git_repack_usage, 0);
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Sep 29, 2025, 23:21 UTC · re: Taylor Blau · lore

Re: [PATCH 12/49] builtin/repack.c: pass both pack_objects args to repack_config

On Sun, Sep 28, 2025 at 06:07:56PM -0400, Taylor Blau wrote:
Show 22 quoted lines
> diff --git a/builtin/repack.c b/builtin/repack.c
> index 6dfcb3327e..af6de8d77a 100644
> --- a/builtin/repack.c
> +++ b/builtin/repack.c
> @@ -1260,6 +1266,7 @@ int cmd_repack(int argc,
>  	size_t midx_pack_names_nr = 0;
>  
>  	/* variables to be filled by option parsing */
> +	struct repack_config_ctx config_ctx;
>  	int delete_redundant = 0;
>  	const char *unpack_unreachable = NULL;
>  	int keep_unreachable = 0;
> @@ -1343,7 +1350,11 @@ int cmd_repack(int argc,
>  
>  	list_objects_filter_init(&po_args.filter_options);
>  
> -	repo_config(repo, repack_config, &cruft_po_args);
> +	memset(&config_ctx, 0, sizeof(config_ctx));
> +	config_ctx.po_args = &po_args;
> +	config_ctx.cruft_po_args = &cruft_po_args;
> +
> +	repo_config(repo, repack_config, &config_ctx);
Nit: same comment here as on a previous commit, we could use a
designated initializer here.
Patrick
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 13/49] repack: move 'delta_base_offset' to 'struct pack_objects_args'

The static variable 'delta_base_offset' determines whether or not we pass the "--delta-base-offset" command-line argument when spawning pack-objects as a child process. Its introduction dates back to when repack was rewritten in C, all the way back in a1bbc6c017 (repack: rewrite the shell script in C, 2013-09-15).

'struct pack_objects_args' was introduced much later on in 4571324b99 (builtin/repack.c: allow configuring cruft pack generation, 2022-05-20), but did not move the 'delta_base_offset' variable.

Since the 'delta_base_offset' is a property of an individual pack-objects command, re-introduce that variable as a member of 'struct pack_objects_args', which will enable further code movement in the subsequent commits.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 11 ++++++-----
 repack.h         |  3 +++
 2 files changed, 9 insertions(+), 5 deletions(-)
Show changes to 2 files +9 −5

builtin/repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index af6de8d77a..f4af830353 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -34,7 +34,6 @@
 #define RETAIN_PACK 2
 
 static int pack_everything;
-static int delta_base_offset = 1;
 static int pack_kept_objects = -1;
 static int write_bitmaps = -1;
 static int use_delta_islands;
@@ -63,9 +62,10 @@ static int repack_config(const char *var, const char *value,
 			 const struct config_context *ctx, void *cb)
 {
 	struct repack_config_ctx *repack_ctx = cb;
+	struct pack_objects_args *po_args = repack_ctx->po_args;
 	struct pack_objects_args *cruft_po_args = repack_ctx->cruft_po_args;
 	if (!strcmp(var, "repack.usedeltabaseoffset")) {
-		delta_base_offset = git_config_bool(var, value);
+		po_args->delta_base_offset = git_config_bool(var, value);
 		return 0;
 	}
 	if (!strcmp(var, "repack.packkeptobjects")) {
@@ -315,7 +315,7 @@ static void prepare_pack_objects(struct child_process *cmd,
 		strvec_push(&cmd->args,  "--local");
 	if (args->quiet)
 		strvec_push(&cmd->args,  "--quiet");
-	if (delta_base_offset)
+	if (args->delta_base_offset)
 		strvec_push(&cmd->args,  "--delta-base-offset");
 	strvec_push(&cmd->args, out);
 	cmd->git_cmd = 1;
@@ -1271,8 +1271,8 @@ int cmd_repack(int argc,
 	const char *unpack_unreachable = NULL;
 	int keep_unreachable = 0;
 	struct string_list keep_pack_list = STRING_LIST_INIT_NODUP;
-	struct pack_objects_args po_args = { 0 };
-	struct pack_objects_args cruft_po_args = { 0 };
+	struct pack_objects_args po_args = PACK_OBJECTS_ARGS_INIT;
+	struct pack_objects_args cruft_po_args = PACK_OBJECTS_ARGS_INIT;
 	int write_midx = 0;
 	const char *cruft_expiration = NULL;
 	const char *expire_to = NULL;
@@ -1567,6 +1567,7 @@ int cmd_repack(int argc,
 
 		cruft_po_args.local = po_args.local;
 		cruft_po_args.quiet = po_args.quiet;
+		cruft_po_args.delta_base_offset = po_args.delta_base_offset;
 
 		ret = write_cruft_pack(&cruft_po_args, packtmp, pack_prefix,
 				       cruft_expiration,
diff --git a/repack.h b/repack.h
index 421d439d5a..12632d7fec 100644
--- a/repack.h
+++ b/repack.h
@@ -15,9 +15,12 @@ struct pack_objects_args {
 	int local;
 	int name_hash_version;
 	int path_walk;
+	int delta_base_offset;
 	struct list_objects_filter_options filter_options;
 };
 
+#define PACK_OBJECTS_ARGS_INIT { .delta_base_offset = 1 }
+
 void pack_objects_args_release(struct pack_objects_args *args);
 
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Jeff King· Oct 10, 2025, 05:54 UTC · re: Taylor Blau · lore

Re: [PATCH 13/49] repack: move 'delta_base_offset' to 'struct pack_objects_args'

On Sun, Sep 28, 2025 at 06:08:00PM -0400, Taylor Blau wrote:
Show 14 quoted lines
> The static variable 'delta_base_offset' determines whether or not we
> pass the "--delta-base-offset" command-line argument when spawning
> pack-objects as a child process. Its introduction dates back to when
> repack was rewritten in C, all the way back in a1bbc6c017 (repack:
> rewrite the shell script in C, 2013-09-15).
> 
> 'struct pack_objects_args' was introduced much later on in 4571324b99
> (builtin/repack.c: allow configuring cruft pack generation, 2022-05-20),
> but did not move the 'delta_base_offset' variable.
> 
> Since the 'delta_base_offset' is a property of an individual
> pack-objects command, re-introduce that variable as a member of 'struct
> pack_objects_args', which will enable further code movement in the
> subsequent commits.

Hmm, OK. It is true that it is a property of a pack-objects invocation, but would we ever want it to be different between the two? I'd think not, and from what I can see:

> +		cruft_po_args.delta_base_offset = po_args.delta_base_offset;

we'd always use the same value in both cases. It kind of makes me wonder if it should be in the repack_ctx variable instead, then, with both pack-objects invocations pulling from a common source. But I could also believe that makes life harder later on, when you want to pass just the pack_objects_args to prepare_pack_objects().

So I dunno. Mainly I am worried about somehow forgetting to set the flag in the right spot, but in practice there really is only place where it needs to get pulled over from po_args to cruft_po_args.

-Peff
Taylor Blau· Oct 10, 2025, 22:54 UTC · re: Jeff King · lore

Re: [PATCH 13/49] repack: move 'delta_base_offset' to 'struct pack_objects_args'

On Fri, Oct 10, 2025 at 01:54:31AM -0400, Jeff King wrote:
Show 28 quoted lines
> On Sun, Sep 28, 2025 at 06:08:00PM -0400, Taylor Blau wrote:
>
> > The static variable 'delta_base_offset' determines whether or not we
> > pass the "--delta-base-offset" command-line argument when spawning
> > pack-objects as a child process. Its introduction dates back to when
> > repack was rewritten in C, all the way back in a1bbc6c017 (repack:
> > rewrite the shell script in C, 2013-09-15).
> >
> > 'struct pack_objects_args' was introduced much later on in 4571324b99
> > (builtin/repack.c: allow configuring cruft pack generation, 2022-05-20),
> > but did not move the 'delta_base_offset' variable.
> >
> > Since the 'delta_base_offset' is a property of an individual
> > pack-objects command, re-introduce that variable as a member of 'struct
> > pack_objects_args', which will enable further code movement in the
> > subsequent commits.
>
> Hmm, OK. It is true that it is a property of a pack-objects invocation,
> but would we ever want it to be different between the two? I'd think
> not, and from what I can see:
>
> > +		cruft_po_args.delta_base_offset = po_args.delta_base_offset;
>
> we'd always use the same value in both cases. It kind of makes me wonder
> if it should be in the repack_ctx variable instead, then, with both
> pack-objects invocations pulling from a common source. But I could also
> believe that makes life harder later on, when you want to pass just the
> pack_objects_args to prepare_pack_objects().

Yeah, we do want to use the same value here in both cases. My thinking when writing this patch was that whether or not we pass the --delta-base-offset command-line argument to pack-objects *is* a property of the pack-objects invocation.

But the fact that we want to use the same value when generating reachable/geometric packs and when generating cruft packs is more of an implementation detail of repack itself. I like that that detail is made explicit in the line above.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 14/49] repack: remove 'prepare_pack_objects' from the builtin

Now that the 'prepare_pack_objects' function no longer refers to external, static variables, move it out to repack.h as generic functionality.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 34 ----------------------------------
 repack.c         | 35 +++++++++++++++++++++++++++++++++++
 repack.h         |  5 +++++
 3 files changed, 40 insertions(+), 34 deletions(-)
Show changes to 3 files +40 −34

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index f4af830353..ff93654cfe 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -288,40 +288,6 @@ static void collect_pack_filenames(struct existing_packs *existing,
 	strbuf_release(&buf);
 }
 
-static void prepare_pack_objects(struct child_process *cmd,
-				 const struct pack_objects_args *args,
-				 const char *out)
-{
-	strvec_push(&cmd->args, "pack-objects");
-	if (args->window)
-		strvec_pushf(&cmd->args, "--window=%s", args->window);
-	if (args->window_memory)
-		strvec_pushf(&cmd->args, "--window-memory=%s", args->window_memory);
-	if (args->depth)
-		strvec_pushf(&cmd->args, "--depth=%s", args->depth);
-	if (args->threads)
-		strvec_pushf(&cmd->args, "--threads=%s", args->threads);
-	if (args->max_pack_size)
-		strvec_pushf(&cmd->args, "--max-pack-size=%lu", args->max_pack_size);
-	if (args->no_reuse_delta)
-		strvec_pushf(&cmd->args, "--no-reuse-delta");
-	if (args->no_reuse_object)
-		strvec_pushf(&cmd->args, "--no-reuse-object");
-	if (args->name_hash_version)
-		strvec_pushf(&cmd->args, "--name-hash-version=%d", args->name_hash_version);
-	if (args->path_walk)
-		strvec_pushf(&cmd->args, "--path-walk");
-	if (args->local)
-		strvec_push(&cmd->args,  "--local");
-	if (args->quiet)
-		strvec_push(&cmd->args,  "--quiet");
-	if (args->delta_base_offset)
-		strvec_push(&cmd->args,  "--delta-base-offset");
-	strvec_push(&cmd->args, out);
-	cmd->git_cmd = 1;
-	cmd->out = -1;
-}
-
 struct write_oid_context {
 	struct child_process *cmd;
 	const struct git_hash_algo *algop;
diff --git a/repack.c b/repack.c
index a1f5b796fb..91b6e1cc09 100644
--- a/repack.c
+++ b/repack.c
@@ -1,5 +1,40 @@
 #include "git-compat-util.h"
 #include "repack.h"
+#include "run-command.h"
+
+void prepare_pack_objects(struct child_process *cmd,
+			  const struct pack_objects_args *args,
+			  const char *out)
+{
+	strvec_push(&cmd->args, "pack-objects");
+	if (args->window)
+		strvec_pushf(&cmd->args, "--window=%s", args->window);
+	if (args->window_memory)
+		strvec_pushf(&cmd->args, "--window-memory=%s", args->window_memory);
+	if (args->depth)
+		strvec_pushf(&cmd->args, "--depth=%s", args->depth);
+	if (args->threads)
+		strvec_pushf(&cmd->args, "--threads=%s", args->threads);
+	if (args->max_pack_size)
+		strvec_pushf(&cmd->args, "--max-pack-size=%lu", args->max_pack_size);
+	if (args->no_reuse_delta)
+		strvec_pushf(&cmd->args, "--no-reuse-delta");
+	if (args->no_reuse_object)
+		strvec_pushf(&cmd->args, "--no-reuse-object");
+	if (args->name_hash_version)
+		strvec_pushf(&cmd->args, "--name-hash-version=%d", args->name_hash_version);
+	if (args->path_walk)
+		strvec_pushf(&cmd->args, "--path-walk");
+	if (args->local)
+		strvec_push(&cmd->args,  "--local");
+	if (args->quiet)
+		strvec_push(&cmd->args,  "--quiet");
+	if (args->delta_base_offset)
+		strvec_push(&cmd->args,  "--delta-base-offset");
+	strvec_push(&cmd->args, out);
+	cmd->git_cmd = 1;
+	cmd->out = -1;
+}
 
 void pack_objects_args_release(struct pack_objects_args *args)
 {
diff --git a/repack.h b/repack.h
index 12632d7fec..3f7ec20735 100644
--- a/repack.h
+++ b/repack.h
@@ -21,6 +21,11 @@ struct pack_objects_args {
 
 #define PACK_OBJECTS_ARGS_INIT { .delta_base_offset = 1 }
 
+struct child_process;
+
+void prepare_pack_objects(struct child_process *cmd,
+			  const struct pack_objects_args *args,
+			  const char *out);
 void pack_objects_args_release(struct pack_objects_args *args);
 
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 15/49] builtin/repack.c: rename many 'struct existing_packs' functions

Rename many of the 'struct existing_packs'-related functions according to the convention introduced in and described by 541204aabe (Documentation: document naming schema for structs and their functions, 2024-07-30).

Note that some functions which operate over an individual entry in the list of existing packs are prefixed with "existing_pack_" instead of the plural form.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 66 +++++++++++++++++++++++++-----------------------
 1 file changed, 34 insertions(+), 32 deletions(-)
Show changes to builtin/repack.c +34 −32
diff --git a/builtin/repack.c b/builtin/repack.c
index ff93654cfe..f82e6c3930 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -121,39 +121,39 @@ struct existing_packs {
 	.cruft_packs = STRING_LIST_INIT_DUP, \
 }
 
-static int has_existing_non_kept_packs(const struct existing_packs *existing)
+static int existing_packs_has_non_kept(const struct existing_packs *existing)
 {
 	return existing->non_kept_packs.nr || existing->cruft_packs.nr;
 }
 
-static void pack_mark_for_deletion(struct string_list_item *item)
+static void existing_pack_mark_for_deletion(struct string_list_item *item)
 {
 	item->util = (void*)((uintptr_t)item->util | DELETE_PACK);
 }
 
-static void pack_unmark_for_deletion(struct string_list_item *item)
+static void existing_pack_unmark_for_deletion(struct string_list_item *item)
 {
 	item->util = (void*)((uintptr_t)item->util & ~DELETE_PACK);
 }
 
-static int pack_is_marked_for_deletion(struct string_list_item *item)
+static int existing_pack_is_marked_for_deletion(struct string_list_item *item)
 {
 	return (uintptr_t)item->util & DELETE_PACK;
 }
 
-static void pack_mark_retained(struct string_list_item *item)
+static void existing_packs_mark_retained(struct string_list_item *item)
 {
 	item->util = (void*)((uintptr_t)item->util | RETAIN_PACK);
 }
 
-static int pack_is_retained(struct string_list_item *item)
+static int existing_pack_is_retained(struct string_list_item *item)
 {
 	return (uintptr_t)item->util & RETAIN_PACK;
 }
 
-static void mark_packs_for_deletion_1(const struct git_hash_algo *algop,
-				      struct string_list *names,
-				      struct string_list *list)
+static void existing_packs_mark_for_deletion_1(const struct git_hash_algo *algop,
+					       struct string_list *names,
+					       struct string_list *list)
 {
 	struct string_list_item *item;
 	const int hexsz = algop->hexsz;
@@ -165,8 +165,8 @@ static void mark_packs_for_deletion_1(const struct git_hash_algo *algop,
 			continue;
 		sha1 = item->string + len - hexsz;
 
-		if (pack_is_retained(item)) {
-			pack_unmark_for_deletion(item);
+		if (existing_pack_is_retained(item)) {
+			existing_pack_unmark_for_deletion(item);
 		} else if (!string_list_has_string(names, sha1)) {
 			/*
 			 * Mark this pack for deletion, which ensures
@@ -175,13 +175,13 @@ static void mark_packs_for_deletion_1(const struct git_hash_algo *algop,
 			 * will actually delete this pack (if `-d` was
 			 * given).
 			 */
-			pack_mark_for_deletion(item);
+			existing_pack_mark_for_deletion(item);
 		}
 	}
 }
 
-static void retain_cruft_pack(struct existing_packs *existing,
-			      struct packed_git *cruft)
+static void existing_packs_retain_cruft(struct existing_packs *existing,
+					struct packed_git *cruft)
 {
 	struct strbuf buf = STRBUF_INIT;
 	struct string_list_item *item;
@@ -193,17 +193,19 @@ static void retain_cruft_pack(struct existing_packs *existing,
 	if (!item)
 		BUG("could not find cruft pack '%s'", pack_basename(cruft));
 
-	pack_mark_retained(item);
+	existing_packs_mark_retained(item);
 	strbuf_release(&buf);
 }
 
-static void mark_packs_for_deletion(struct existing_packs *existing,
-				    struct string_list *names)
+static void existing_packs_mark_for_deletion(struct existing_packs *existing,
+					     struct string_list *names)
 
 {
 	const struct git_hash_algo *algop = existing->repo->hash_algo;
-	mark_packs_for_deletion_1(algop, names, &existing->non_kept_packs);
-	mark_packs_for_deletion_1(algop, names, &existing->cruft_packs);
+	existing_packs_mark_for_deletion_1(algop, names,
+					   &existing->non_kept_packs);
+	existing_packs_mark_for_deletion_1(algop, names,
+					   &existing->cruft_packs);
 }
 
 static void remove_redundant_pack(struct repository *repo,
@@ -225,13 +227,13 @@ static void remove_redundant_packs_1(struct repository *repo,
 {
 	struct string_list_item *item;
 	for_each_string_list_item(item, packs) {
-		if (!pack_is_marked_for_deletion(item))
+		if (!existing_pack_is_marked_for_deletion(item))
 			continue;
 		remove_redundant_pack(repo, packdir, item->string);
 	}
 }
 
-static void remove_redundant_existing_packs(struct existing_packs *existing)
+static void existing_packs_remove_redundant(struct existing_packs *existing)
 {
 	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs);
 	remove_redundant_packs_1(existing->repo, &existing->cruft_packs);
@@ -250,7 +252,7 @@ static void existing_packs_release(struct existing_packs *existing)
  * .keep file or not.  Packs without a .keep file are not to be kept
  * if we are going to pack everything into one file.
  */
-static void collect_pack_filenames(struct existing_packs *existing,
+static void existing_packs_collect(struct existing_packs *existing,
 				   const struct string_list *extra_keep)
 {
 	struct packfile_store *packs = existing->repo->objects->packfiles;
@@ -721,7 +723,7 @@ static int midx_has_unknown_packs(char **midx_pack_names,
 
 			item = string_list_lookup(&existing->non_kept_packs,
 						  pack_name);
-			if (item && !pack_is_marked_for_deletion(item))
+			if (item && !existing_pack_is_marked_for_deletion(item))
 				continue;
 		}
 
@@ -851,7 +853,7 @@ static void midx_included_packs(struct string_list *include,
 		}
 	} else {
 		for_each_string_list_item(item, &existing->non_kept_packs) {
-			if (pack_is_marked_for_deletion(item))
+			if (existing_pack_is_marked_for_deletion(item))
 				continue;
 
 			strbuf_reset(&buf);
@@ -888,10 +890,10 @@ static void midx_included_packs(struct string_list *include,
 			 * --geometric case, but doing so is unnecessary
 			 *  since no packs are marked as pending
 			 *  deletion (since we only call
-			 *  `mark_packs_for_deletion()` when doing an
-			 *  all-into-one repack).
+			 *  `existing_packs_mark_for_deletion()` when
+			 *  doing an all-into-one repack).
 			 */
-			if (pack_is_marked_for_deletion(item))
+			if (existing_pack_is_marked_for_deletion(item))
 				continue;
 
 			strbuf_reset(&buf);
@@ -1128,7 +1130,7 @@ static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
 		if (p->pack_size < combine_cruft_below_size) {
 			fprintf(in, "-%s\n", pack_basename(p));
 		} else {
-			retain_cruft_pack(existing, p);
+			existing_packs_retain_cruft(existing, p);
 			fprintf(in, "%s\n", pack_basename(p));
 		}
 	}
@@ -1382,7 +1384,7 @@ int cmd_repack(int argc,
 	packtmp = mkpathdup("%s/%s", packdir, packtmp_name);
 
 	existing.repo = repo;
-	collect_pack_filenames(&existing, &keep_pack_list);
+	existing_packs_collect(&existing, &keep_pack_list);
 
 	if (geometry.split_factor) {
 		if (pack_everything)
@@ -1431,7 +1433,7 @@ int cmd_repack(int argc,
 	if (pack_everything & ALL_INTO_ONE) {
 		repack_promisor_objects(repo, &po_args, &names);
 
-		if (has_existing_non_kept_packs(&existing) &&
+		if (existing_packs_has_non_kept(&existing) &&
 		    delete_redundant &&
 		    !(pack_everything & PACK_CRUFT)) {
 			for_each_string_list_item(item, &names) {
@@ -1647,7 +1649,7 @@ int cmd_repack(int argc,
 	/* End of pack replacement. */
 
 	if (delete_redundant && pack_everything & ALL_INTO_ONE)
-		mark_packs_for_deletion(&existing, &names);
+		existing_packs_mark_for_deletion(&existing, &names);
 
 	if (write_midx) {
 		struct string_list include = STRING_LIST_INIT_DUP;
@@ -1671,7 +1673,7 @@ int cmd_repack(int argc,
 
 	if (delete_redundant) {
 		int opts = 0;
-		remove_redundant_existing_packs(&existing);
+		existing_packs_remove_redundant(&existing);
 
 		if (geometry.split_factor)
 			geometry_remove_redundant_packs(&geometry, &names,
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 16/49] repack: remove 'remove_redundant_pack' from the builtin

Extract "remove_redundant_pack()" as generic repack-related functionality by moving its implementation to the repack.[ch] compilation unit.

This is a prerequisite to moving the "existing_packs" API, which is one of the callers of this function. (The remaining caller in the pack geometry code will eventually move to its own compilation unit as well, and will likewise rely on this function.)

While moving it over, prefix the function name with "repack_" to indicate that it belongs to the repack-subsystem.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 18 ++----------------
 repack.c         | 18 ++++++++++++++++++
 repack.h         |  3 +++
 3 files changed, 23 insertions(+), 16 deletions(-)
Show changes to 3 files +23 −16

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index f82e6c3930..31137cf711 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -208,20 +208,6 @@ static void existing_packs_mark_for_deletion(struct existing_packs *existing,
 					   &existing->cruft_packs);
 }
 
-static void remove_redundant_pack(struct repository *repo,
-				  const char *dir_name, const char *base_name)
-{
-	struct strbuf buf = STRBUF_INIT;
-	struct odb_source *source = repo->objects->sources;
-	struct multi_pack_index *m = get_multi_pack_index(source);
-	strbuf_addf(&buf, "%s.pack", base_name);
-	if (m && source->local && midx_contains_pack(m, buf.buf))
-		clear_midx_file(repo);
-	strbuf_insertf(&buf, 0, "%s/", dir_name);
-	unlink_pack_path(buf.buf, 1);
-	strbuf_release(&buf);
-}
-
 static void remove_redundant_packs_1(struct repository *repo,
 				     struct string_list *packs)
 {
@@ -229,7 +215,7 @@ static void remove_redundant_packs_1(struct repository *repo,
 	for_each_string_list_item(item, packs) {
 		if (!existing_pack_is_marked_for_deletion(item))
 			continue;
-		remove_redundant_pack(repo, packdir, item->string);
+		repack_remove_redundant_pack(repo, packdir, item->string);
 	}
 }
 
@@ -652,7 +638,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
 		    (string_list_has_string(&existing->kept_packs, buf.buf)))
 			continue;
 
-		remove_redundant_pack(existing->repo, packdir, buf.buf);
+		repack_remove_redundant_pack(existing->repo, packdir, buf.buf);
 	}
 
 	strbuf_release(&buf);
diff --git a/repack.c b/repack.c
index 91b6e1cc09..3aaa351b5b 100644
--- a/repack.c
+++ b/repack.c
@@ -1,5 +1,9 @@
 #include "git-compat-util.h"
+#include "midx.h"
+#include "odb.h"
+#include "packfile.h"
 #include "repack.h"
+#include "repository.h"
 #include "run-command.h"
 
 void prepare_pack_objects(struct child_process *cmd,
@@ -44,3 +48,17 @@ void pack_objects_args_release(struct pack_objects_args *args)
 	free(args->threads);
 	list_objects_filter_release(&args->filter_options);
 }
+
+void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
+				  const char *base_name)
+{
+	struct strbuf buf = STRBUF_INIT;
+	struct odb_source *source = repo->objects->sources;
+	struct multi_pack_index *m = get_multi_pack_index(source);
+	strbuf_addf(&buf, "%s.pack", base_name);
+	if (m && source->local && midx_contains_pack(m, buf.buf))
+		clear_midx_file(repo);
+	strbuf_insertf(&buf, 0, "%s/", dir_name);
+	unlink_pack_path(buf.buf, 1);
+	strbuf_release(&buf);
+}
diff --git a/repack.h b/repack.h
index 3f7ec20735..a62bfa2ff9 100644
--- a/repack.h
+++ b/repack.h
@@ -28,4 +28,7 @@ void prepare_pack_objects(struct child_process *cmd,
 			  const char *out);
 void pack_objects_args_release(struct pack_objects_args *args);
 
+void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
+				  const char *base_name);
+
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 17/49] builtin/repack.c: pass "packdir" when removing packs

builtin/repack.c defines a static "packdir" to instruct pack-objects on where to write any new packfiles. This is also the directory scanned when removing any packfiles which were made redundant by the latest repack.

Prepare to move the "existing_packs_remove_redundant" function to its own compilation unit by passing in this information as a parameter to that function.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
Show changes to builtin/repack.c +9 −5
diff --git a/builtin/repack.c b/builtin/repack.c
index 31137cf711..c5a88eda12 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -209,7 +209,8 @@ static void existing_packs_mark_for_deletion(struct existing_packs *existing,
 }
 
 static void remove_redundant_packs_1(struct repository *repo,
-				     struct string_list *packs)
+				     struct string_list *packs,
+				     const char *packdir)
 {
 	struct string_list_item *item;
 	for_each_string_list_item(item, packs) {
@@ -219,10 +220,13 @@ static void remove_redundant_packs_1(struct repository *repo,
 	}
 }
 
-static void existing_packs_remove_redundant(struct existing_packs *existing)
+static void existing_packs_remove_redundant(struct existing_packs *existing,
+					    const char *packdir)
 {
-	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs);
-	remove_redundant_packs_1(existing->repo, &existing->cruft_packs);
+	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs,
+				 packdir);
+	remove_redundant_packs_1(existing->repo, &existing->cruft_packs,
+				 packdir);
 }
 
 static void existing_packs_release(struct existing_packs *existing)
@@ -1659,7 +1663,7 @@ int cmd_repack(int argc,
 
 	if (delete_redundant) {
 		int opts = 0;
-		existing_packs_remove_redundant(&existing);
+		existing_packs_remove_redundant(&existing, packdir);
 
 		if (geometry.split_factor)
 			geometry_remove_redundant_packs(&geometry, &names,
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Sep 29, 2025, 23:21 UTC · re: Taylor Blau · lore

Re: [PATCH 17/49] builtin/repack.c: pass "packdir" when removing packs

On Sun, Sep 28, 2025 at 06:08:16PM -0400, Taylor Blau wrote:
Show 8 quoted lines
> builtin/repack.c defines a static "packdir" to instruct pack-objects on
> where to write any new packfiles. This is also the directory scanned
> when removing any packfiles which were made redundant by the latest
> repack.
> 
> Prepare to move the "existing_packs_remove_redundant" function to its
> own compilation unit by passing in this information as a parameter to
> that function.

I would have expected that we also get rid of that static variable either in this commit or a follow-up. Is there any reason why you skipped over this? If so, it would be great to highlight in the commit message.

Patrick
Taylor Blau· Oct 7, 2025, 20:23 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 17/49] builtin/repack.c: pass "packdir" when removing packs

On Tue, Sep 30, 2025 at 01:21:34AM +0200, Patrick Steinhardt wrote:
Show 14 quoted lines
> On Sun, Sep 28, 2025 at 06:08:16PM -0400, Taylor Blau wrote:
> > builtin/repack.c defines a static "packdir" to instruct pack-objects on
> > where to write any new packfiles. This is also the directory scanned
> > when removing any packfiles which were made redundant by the latest
> > repack.
> >
> > Prepare to move the "existing_packs_remove_redundant" function to its
> > own compilation unit by passing in this information as a parameter to
> > that function.
>
> I would have expected that we also get rid of that static variable
> either in this commit or a follow-up. Is there any reason why you
> skipped over this? If so, it would be great to highlight in the commit
> message.

We can't get rid of "packdir" quite yet, because it's used in other functions (which get a similar treatment as this one does in later patches).

We could (and probably should) drop this, packtmp_name, and packtmp as static variables within builtin/repack.c. We could do it as a fiftieth patch in this series like so:

--- 8< ---
Show changes to builtin/repack.c +1 −1
diff --git a/builtin/repack.c b/builtin/repack.c
index ad60c4290d..565538bb24 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -27,7 +27,6 @@ static int pack_everything;
 static int write_bitmaps = -1;
 static int use_delta_islands;
 static int run_update_server_info = 1;
-static char *packdir, *packtmp_name, *packtmp;
 static int midx_must_contain_cruft = 1;

 static const char *const git_repack_usage[] = {
@@ -111,6 +110,7 @@ int cmd_repack(int argc,
 	struct write_pack_opts opts = { 0 };
 	int i, ret;
 	int show_progress;
+	char *packdir, *packtmp_name, *packtmp;

 	/* variables to be filled by option parsing */
 	struct repack_config_ctx config_ctx;
--- >8 ---

But I'd like to err on the side of post-merge followups here to avoid
dumping 49+ more patches on the list. Of course, if there are
showstoppers or other objections, I can easily reroll the series, but
I'd rather avoid doing so on a series this long.

Thanks,
Taylor
Jeff King· Oct 10, 2025, 06:04 UTC · re: Taylor Blau · lore

Re: [PATCH 17/49] builtin/repack.c: pass "packdir" when removing packs

On Sun, Sep 28, 2025 at 06:08:16PM -0400, Taylor Blau wrote:
Show 8 quoted lines
> builtin/repack.c defines a static "packdir" to instruct pack-objects on
> where to write any new packfiles. This is also the directory scanned
> when removing any packfiles which were made redundant by the latest
> repack.
> 
> Prepare to move the "existing_packs_remove_redundant" function to its
> own compilation unit by passing in this information as a parameter to
> that function.

OK. I don't think there's anything wrong here, but it feels like we ought to be able to pull this from the existing "struct repository" context that's in the existing_packs struct.

I had to dig a little to find it, but I think that always comes from:
  packdir = mkpathdup("%s/pack", repo_get_object_directory(repo));

in cmd_repack(). So we'd have to recreate that string ourselves. It feels like object_database (or I guess odb_source) should have an easy way to get the pack directory, but it doesn't seem to.

So I dunno. I think it is mostly academic, as you are just moving around code here. But as an API, the separate "packdir" argument makes me wonder if I can pass anything I like to these functions. But really, it should match the packdir used to create the lists in the existing_packs struct. So it's kind of a brittle API. I'm not sure how much we should care for this series, as it is mostly about moving code around, and not providing some kind of flexible repack API.

-Peff
Taylor Blau· Oct 10, 2025, 22:56 UTC · re: Jeff King · lore

Re: [PATCH 17/49] builtin/repack.c: pass "packdir" when removing packs

On Fri, Oct 10, 2025 at 02:04:01AM -0400, Jeff King wrote:
Show 30 quoted lines
> On Sun, Sep 28, 2025 at 06:08:16PM -0400, Taylor Blau wrote:
>
> > builtin/repack.c defines a static "packdir" to instruct pack-objects on
> > where to write any new packfiles. This is also the directory scanned
> > when removing any packfiles which were made redundant by the latest
> > repack.
> >
> > Prepare to move the "existing_packs_remove_redundant" function to its
> > own compilation unit by passing in this information as a parameter to
> > that function.
>
> OK. I don't think there's anything wrong here, but it feels like we
> ought to be able to pull this from the existing "struct repository"
> context that's in the existing_packs struct.
>
> I had to dig a little to find it, but I think that always comes from:
>
>   packdir = mkpathdup("%s/pack", repo_get_object_directory(repo));
>
> in cmd_repack(). So we'd have to recreate that string ourselves. It
> feels like object_database (or I guess odb_source) should have an easy
> way to get the pack directory, but it doesn't seem to.
>
> So I dunno. I think it is mostly academic, as you are just moving around
> code here. But as an API, the separate "packdir" argument makes me
> wonder if I can pass anything I like to these functions. But really, it
> should match the packdir used to create the lists in the existing_packs
> struct. So it's kind of a brittle API. I'm not sure how much we should
> care for this series, as it is mostly about moving code around, and not
> providing some kind of flexible repack API.
Oh, it's extremely brittle, I agree ;-).

I agree with your assessment here that packdir can easily be derived from the repository object, and that doing so would result in a cleaner, less brittle API. I'd like to make those changes, but again would prefer to do so on top of this series, since I want to change as few things as possible here.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 18/49] builtin/repack.c: avoid unnecessary numeric casts in existing_packs

There are a couple of spots that cause warnings within the existing_packs API without DISABLE_SIGN_COMPARE_WARNINGS under DEVELOPER=1 mode.

In both cases, we have int values that are being compared against size_t ones. Neither of these two cases are incorrect, and the cast is completely OK in practice. But both are unnecessary, since:

 - in existing_packs_mark_for_deletion_1(), 'hexsz' should be defined as
   a size_t anyway, since algop->hexsz is.
 - in existing_packs_collect(), 'i' should be defined as a size_t since
   it is counting up to the value of a string_list's 'nr' field.

(This patch is a little bit of noise, but I would rather see us squelch these warnings ahead of moving the existing_packs API into a separate compilation unit to avoid having to define DISABLE_SIGN_COMPARE_WARNINGS in repack.c.)

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +2 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index c5a88eda12..e13943b637 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -156,7 +156,7 @@ static void existing_packs_mark_for_deletion_1(const struct git_hash_algo *algop
 					       struct string_list *list)
 {
 	struct string_list_item *item;
-	const int hexsz = algop->hexsz;
+	const size_t hexsz = algop->hexsz;
 
 	for_each_string_list_item(item, list) {
 		char *sha1;
@@ -250,7 +250,7 @@ static void existing_packs_collect(struct existing_packs *existing,
 	struct strbuf buf = STRBUF_INIT;
 
 	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
-		int i;
+		size_t i;
 		const char *base;
 
 		if (!p->pack_local)
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 19/49] repack: remove 'existing_packs' API from the builtin

The repack builtin defines an API for keeping track of which packs were found in the repository at the beginning of the repack operation. This is used to classify what state a pack was in (kept, non-kept, or cruft), and is also used to mark which packs to delete (or keep) at the end of a repack operation.

Now that the prerequisite refactoring is complete, this API is isolated enough that it can be moved out to repack.ch and removed from the builtin entirely.

As a result, some of its functions become static within repack.c, cleaning up the visible API.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 173 -----------------------------------------------
 repack.c         | 157 ++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  35 ++++++++++
 3 files changed, 192 insertions(+), 173 deletions(-)
Show changes to 3 files +192 −173

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index e13943b637..a168c88791 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -3,7 +3,6 @@
 
 #include "builtin.h"
 #include "config.h"
-#include "dir.h"
 #include "environment.h"
 #include "gettext.h"
 #include "hex.h"
@@ -108,178 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-struct existing_packs {
-	struct repository *repo;
-	struct string_list kept_packs;
-	struct string_list non_kept_packs;
-	struct string_list cruft_packs;
-};
-
-#define EXISTING_PACKS_INIT { \
-	.kept_packs = STRING_LIST_INIT_DUP, \
-	.non_kept_packs = STRING_LIST_INIT_DUP, \
-	.cruft_packs = STRING_LIST_INIT_DUP, \
-}
-
-static int existing_packs_has_non_kept(const struct existing_packs *existing)
-{
-	return existing->non_kept_packs.nr || existing->cruft_packs.nr;
-}
-
-static void existing_pack_mark_for_deletion(struct string_list_item *item)
-{
-	item->util = (void*)((uintptr_t)item->util | DELETE_PACK);
-}
-
-static void existing_pack_unmark_for_deletion(struct string_list_item *item)
-{
-	item->util = (void*)((uintptr_t)item->util & ~DELETE_PACK);
-}
-
-static int existing_pack_is_marked_for_deletion(struct string_list_item *item)
-{
-	return (uintptr_t)item->util & DELETE_PACK;
-}
-
-static void existing_packs_mark_retained(struct string_list_item *item)
-{
-	item->util = (void*)((uintptr_t)item->util | RETAIN_PACK);
-}
-
-static int existing_pack_is_retained(struct string_list_item *item)
-{
-	return (uintptr_t)item->util & RETAIN_PACK;
-}
-
-static void existing_packs_mark_for_deletion_1(const struct git_hash_algo *algop,
-					       struct string_list *names,
-					       struct string_list *list)
-{
-	struct string_list_item *item;
-	const size_t hexsz = algop->hexsz;
-
-	for_each_string_list_item(item, list) {
-		char *sha1;
-		size_t len = strlen(item->string);
-		if (len < hexsz)
-			continue;
-		sha1 = item->string + len - hexsz;
-
-		if (existing_pack_is_retained(item)) {
-			existing_pack_unmark_for_deletion(item);
-		} else if (!string_list_has_string(names, sha1)) {
-			/*
-			 * Mark this pack for deletion, which ensures
-			 * that this pack won't be included in a MIDX
-			 * (if `--write-midx` was given) and that we
-			 * will actually delete this pack (if `-d` was
-			 * given).
-			 */
-			existing_pack_mark_for_deletion(item);
-		}
-	}
-}
-
-static void existing_packs_retain_cruft(struct existing_packs *existing,
-					struct packed_git *cruft)
-{
-	struct strbuf buf = STRBUF_INIT;
-	struct string_list_item *item;
-
-	strbuf_addstr(&buf, pack_basename(cruft));
-	strbuf_strip_suffix(&buf, ".pack");
-
-	item = string_list_lookup(&existing->cruft_packs, buf.buf);
-	if (!item)
-		BUG("could not find cruft pack '%s'", pack_basename(cruft));
-
-	existing_packs_mark_retained(item);
-	strbuf_release(&buf);
-}
-
-static void existing_packs_mark_for_deletion(struct existing_packs *existing,
-					     struct string_list *names)
-
-{
-	const struct git_hash_algo *algop = existing->repo->hash_algo;
-	existing_packs_mark_for_deletion_1(algop, names,
-					   &existing->non_kept_packs);
-	existing_packs_mark_for_deletion_1(algop, names,
-					   &existing->cruft_packs);
-}
-
-static void remove_redundant_packs_1(struct repository *repo,
-				     struct string_list *packs,
-				     const char *packdir)
-{
-	struct string_list_item *item;
-	for_each_string_list_item(item, packs) {
-		if (!existing_pack_is_marked_for_deletion(item))
-			continue;
-		repack_remove_redundant_pack(repo, packdir, item->string);
-	}
-}
-
-static void existing_packs_remove_redundant(struct existing_packs *existing,
-					    const char *packdir)
-{
-	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs,
-				 packdir);
-	remove_redundant_packs_1(existing->repo, &existing->cruft_packs,
-				 packdir);
-}
-
-static void existing_packs_release(struct existing_packs *existing)
-{
-	string_list_clear(&existing->kept_packs, 0);
-	string_list_clear(&existing->non_kept_packs, 0);
-	string_list_clear(&existing->cruft_packs, 0);
-}
-
-/*
- * Adds all packs hex strings (pack-$HASH) to either packs->non_kept
- * or packs->kept based on whether each pack has a corresponding
- * .keep file or not.  Packs without a .keep file are not to be kept
- * if we are going to pack everything into one file.
- */
-static void existing_packs_collect(struct existing_packs *existing,
-				   const struct string_list *extra_keep)
-{
-	struct packfile_store *packs = existing->repo->objects->packfiles;
-	struct packed_git *p;
-	struct strbuf buf = STRBUF_INIT;
-
-	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
-		size_t i;
-		const char *base;
-
-		if (!p->pack_local)
-			continue;
-
-		base = pack_basename(p);
-
-		for (i = 0; i < extra_keep->nr; i++)
-			if (!fspathcmp(base, extra_keep->items[i].string))
-				break;
-
-		strbuf_reset(&buf);
-		strbuf_addstr(&buf, base);
-		strbuf_strip_suffix(&buf, ".pack");
-
-		if ((extra_keep->nr > 0 && i < extra_keep->nr) || p->pack_keep)
-			string_list_append(&existing->kept_packs, buf.buf);
-		else if (p->is_cruft)
-			string_list_append(&existing->cruft_packs, buf.buf);
-		else
-			string_list_append(&existing->non_kept_packs, buf.buf);
-	}
-
-	string_list_sort(&existing->kept_packs);
-	string_list_sort(&existing->non_kept_packs);
-	string_list_sort(&existing->cruft_packs);
-	strbuf_release(&buf);
-}
-
 struct write_oid_context {
 	struct child_process *cmd;
 	const struct git_hash_algo *algop;
diff --git a/repack.c b/repack.c
index 3aaa351b5b..9182e1c50b 100644
--- a/repack.c
+++ b/repack.c
@@ -1,4 +1,5 @@
 #include "git-compat-util.h"
+#include "dir.h"
 #include "midx.h"
 #include "odb.h"
 #include "packfile.h"
@@ -62,3 +63,159 @@ void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 	unlink_pack_path(buf.buf, 1);
 	strbuf_release(&buf);
 }
+
+#define DELETE_PACK 1
+#define RETAIN_PACK 2
+
+void existing_packs_collect(struct existing_packs *existing,
+			    const struct string_list *extra_keep)
+{
+	struct packfile_store *packs = existing->repo->objects->packfiles;
+	struct packed_git *p;
+	struct strbuf buf = STRBUF_INIT;
+
+	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
+		size_t i;
+		const char *base;
+
+		if (!p->pack_local)
+			continue;
+
+		base = pack_basename(p);
+
+		for (i = 0; i < extra_keep->nr; i++)
+			if (!fspathcmp(base, extra_keep->items[i].string))
+				break;
+
+		strbuf_reset(&buf);
+		strbuf_addstr(&buf, base);
+		strbuf_strip_suffix(&buf, ".pack");
+
+		if ((extra_keep->nr > 0 && i < extra_keep->nr) || p->pack_keep)
+			string_list_append(&existing->kept_packs, buf.buf);
+		else if (p->is_cruft)
+			string_list_append(&existing->cruft_packs, buf.buf);
+		else
+			string_list_append(&existing->non_kept_packs, buf.buf);
+	}
+
+	string_list_sort(&existing->kept_packs);
+	string_list_sort(&existing->non_kept_packs);
+	string_list_sort(&existing->cruft_packs);
+	strbuf_release(&buf);
+}
+
+int existing_packs_has_non_kept(const struct existing_packs *existing)
+{
+	return existing->non_kept_packs.nr || existing->cruft_packs.nr;
+}
+
+static void existing_pack_mark_for_deletion(struct string_list_item *item)
+{
+	item->util = (void*)((uintptr_t)item->util | DELETE_PACK);
+}
+
+static void existing_pack_unmark_for_deletion(struct string_list_item *item)
+{
+	item->util = (void*)((uintptr_t)item->util & ~DELETE_PACK);
+}
+
+int existing_pack_is_marked_for_deletion(struct string_list_item *item)
+{
+	return (uintptr_t)item->util & DELETE_PACK;
+}
+
+static void existing_packs_mark_retained(struct string_list_item *item)
+{
+	item->util = (void*)((uintptr_t)item->util | RETAIN_PACK);
+}
+
+static int existing_pack_is_retained(struct string_list_item *item)
+{
+	return (uintptr_t)item->util & RETAIN_PACK;
+}
+
+static void existing_packs_mark_for_deletion_1(const struct git_hash_algo *algop,
+					       struct string_list *names,
+					       struct string_list *list)
+{
+	struct string_list_item *item;
+	const size_t hexsz = algop->hexsz;
+
+	for_each_string_list_item(item, list) {
+		char *sha1;
+		size_t len = strlen(item->string);
+		if (len < hexsz)
+			continue;
+		sha1 = item->string + len - hexsz;
+
+		if (existing_pack_is_retained(item)) {
+			existing_pack_unmark_for_deletion(item);
+		} else if (!string_list_has_string(names, sha1)) {
+			/*
+			 * Mark this pack for deletion, which ensures
+			 * that this pack won't be included in a MIDX
+			 * (if `--write-midx` was given) and that we
+			 * will actually delete this pack (if `-d` was
+			 * given).
+			 */
+			existing_pack_mark_for_deletion(item);
+		}
+	}
+}
+
+void existing_packs_retain_cruft(struct existing_packs *existing,
+				 struct packed_git *cruft)
+{
+	struct strbuf buf = STRBUF_INIT;
+	struct string_list_item *item;
+
+	strbuf_addstr(&buf, pack_basename(cruft));
+	strbuf_strip_suffix(&buf, ".pack");
+
+	item = string_list_lookup(&existing->cruft_packs, buf.buf);
+	if (!item)
+		BUG("could not find cruft pack '%s'", pack_basename(cruft));
+
+	existing_packs_mark_retained(item);
+	strbuf_release(&buf);
+}
+
+void existing_packs_mark_for_deletion(struct existing_packs *existing,
+				      struct string_list *names)
+
+{
+	const struct git_hash_algo *algop = existing->repo->hash_algo;
+	existing_packs_mark_for_deletion_1(algop, names,
+					   &existing->non_kept_packs);
+	existing_packs_mark_for_deletion_1(algop, names,
+					   &existing->cruft_packs);
+}
+
+static void remove_redundant_packs_1(struct repository *repo,
+				     struct string_list *packs,
+				     const char *packdir)
+{
+	struct string_list_item *item;
+	for_each_string_list_item(item, packs) {
+		if (!existing_pack_is_marked_for_deletion(item))
+			continue;
+		repack_remove_redundant_pack(repo, packdir, item->string);
+	}
+}
+
+void existing_packs_remove_redundant(struct existing_packs *existing,
+				     const char *packdir)
+{
+	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs,
+				 packdir);
+	remove_redundant_packs_1(existing->repo, &existing->cruft_packs,
+				 packdir);
+}
+
+void existing_packs_release(struct existing_packs *existing)
+{
+	string_list_clear(&existing->kept_packs, 0);
+	string_list_clear(&existing->non_kept_packs, 0);
+	string_list_clear(&existing->cruft_packs, 0);
+}
diff --git a/repack.h b/repack.h
index a62bfa2ff9..19796e2243 100644
--- a/repack.h
+++ b/repack.h
@@ -2,6 +2,7 @@
 #define REPACK_H
 
 #include "list-objects-filter-options.h"
+#include "string-list.h"
 
 struct pack_objects_args {
 	char *window;
@@ -31,4 +32,38 @@ void pack_objects_args_release(struct pack_objects_args *args);
 void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 				  const char *base_name);
 
+struct repository;
+struct packed_git;
+
+struct existing_packs {
+	struct repository *repo;
+	struct string_list kept_packs;
+	struct string_list non_kept_packs;
+	struct string_list cruft_packs;
+};
+
+#define EXISTING_PACKS_INIT { \
+	.kept_packs = STRING_LIST_INIT_DUP, \
+	.non_kept_packs = STRING_LIST_INIT_DUP, \
+	.cruft_packs = STRING_LIST_INIT_DUP, \
+}
+
+/*
+ * Adds all packs hex strings (pack-$HASH) to either packs->non_kept
+ * or packs->kept based on whether each pack has a corresponding
+ * .keep file or not.  Packs without a .keep file are not to be kept
+ * if we are going to pack everything into one file.
+ */
+void existing_packs_collect(struct existing_packs *existing,
+			    const struct string_list *extra_keep);
+int existing_packs_has_non_kept(const struct existing_packs *existing);
+int existing_pack_is_marked_for_deletion(struct string_list_item *item);
+void existing_packs_retain_cruft(struct existing_packs *existing,
+				 struct packed_git *cruft);
+void existing_packs_mark_for_deletion(struct existing_packs *existing,
+				      struct string_list *names);
+void existing_packs_remove_redundant(struct existing_packs *existing,
+				     const char *packdir);
+void existing_packs_release(struct existing_packs *existing);
+
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Sep 29, 2025, 23:21 UTC · re: Taylor Blau · lore

Re: [PATCH 19/49] repack: remove 'existing_packs' API from the builtin

On Sun, Sep 28, 2025 at 06:08:23PM -0400, Taylor Blau wrote:
Show 8 quoted lines
> The repack builtin defines an API for keeping track of which packs
> were found in the repository at the beginning of the repack operation.
> This is used to classify what state a pack was in (kept, non-kept, or
> cruft), and is also used to mark which packs to delete (or keep) at the
> end of a repack operation.
> 
> Now that the prerequisite refactoring is complete, this API is isolated
> enough that it can be moved out to repack.ch and removed from the

Tiniest nit, sorry: I found the spelling of "repack.ch" a bit confusing at first, as it looked like a typo. Maybe say "repack.{c,h}" or "repack.[ch]"?

Patrick
Taylor Blau· Oct 7, 2025, 20:24 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 19/49] repack: remove 'existing_packs' API from the builtin

On Tue, Sep 30, 2025 at 01:21:40AM +0200, Patrick Steinhardt wrote:
Show 13 quoted lines
> On Sun, Sep 28, 2025 at 06:08:23PM -0400, Taylor Blau wrote:
> > The repack builtin defines an API for keeping track of which packs
> > were found in the repository at the beginning of the repack operation.
> > This is used to classify what state a pack was in (kept, non-kept, or
> > cruft), and is also used to mark which packs to delete (or keep) at the
> > end of a repack operation.
> >
> > Now that the prerequisite refactoring is complete, this API is isolated
> > enough that it can be moved out to repack.ch and removed from the
>
> Tiniest nit, sorry: I found the spelling of "repack.ch" a bit confusing
> at first, as it looked like a typo. Maybe say "repack.{c,h}" or
> "repack.[ch]"?

Ah, sorry, I usually try and write ".[ch]", but I must have missed it here. I'll adjust it locally and include it in the next round (if any).

Thanks, Taylor

Jeff King· Oct 10, 2025, 06:08 UTC · re: Taylor Blau · lore

Re: [PATCH 19/49] repack: remove 'existing_packs' API from the builtin

On Sun, Sep 28, 2025 at 06:08:23PM -0400, Taylor Blau wrote:
Show 23 quoted lines
> --- a/repack.h
> +++ b/repack.h
> @@ -2,6 +2,7 @@
>  #define REPACK_H
>  
>  #include "list-objects-filter-options.h"
> +#include "string-list.h"
>  
>  struct pack_objects_args {
>  	char *window;
> @@ -31,4 +32,38 @@ void pack_objects_args_release(struct pack_objects_args *args);
>  void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
>  				  const char *base_name);
>  
> +struct repository;
> +struct packed_git;
> +
> +struct existing_packs {
> +	struct repository *repo;
> +	struct string_list kept_packs;
> +	struct string_list non_kept_packs;
> +	struct string_list cruft_packs;
> +};

Earlier, I saw remove_redundant_pack() got namespaced with a "repack_" prefix. Which I think makes sense, since it's a pretty generic name and the symbol now has global visibility. But none of this existing_packs stuff got the same treatment.

I don't know how much it matters. I certainly have no plans to call something existing_packs elsewhere in the code, but I do wonder if the naming should make it more clear this is all repack-related.

-Peff
Taylor Blau· Oct 10, 2025, 22:57 UTC · re: Jeff King · lore

Re: [PATCH 19/49] repack: remove 'existing_packs' API from the builtin

On Fri, Oct 10, 2025 at 02:08:36AM -0400, Jeff King wrote:
Show 34 quoted lines
> On Sun, Sep 28, 2025 at 06:08:23PM -0400, Taylor Blau wrote:
>
> > --- a/repack.h
> > +++ b/repack.h
> > @@ -2,6 +2,7 @@
> >  #define REPACK_H
> >
> >  #include "list-objects-filter-options.h"
> > +#include "string-list.h"
> >
> >  struct pack_objects_args {
> >  	char *window;
> > @@ -31,4 +32,38 @@ void pack_objects_args_release(struct pack_objects_args *args);
> >  void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
> >  				  const char *base_name);
> >
> > +struct repository;
> > +struct packed_git;
> > +
> > +struct existing_packs {
> > +	struct repository *repo;
> > +	struct string_list kept_packs;
> > +	struct string_list non_kept_packs;
> > +	struct string_list cruft_packs;
> > +};
>
> Earlier, I saw remove_redundant_pack() got namespaced with a "repack_"
> prefix. Which I think makes sense, since it's a pretty generic name and
> the symbol now has global visibility. But none of this existing_packs
> stuff got the same treatment.
>
> I don't know how much it matters. I certainly have no plans to call
> something existing_packs elsewhere in the code, but I do wonder if the
> naming should make it more clear this is all repack-related.

I think this is a similar case as `struct pack_objects_args`. We could prefix this as `repack_existing_packs`, but it seems a bit verbose, and existing_packs is a semi-useful API by itself.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 20/49] builtin/repack.c: rename "struct generated_pack_data"

The name "generated_pack_data" is somewhat redundant, since the contents of the struct *is* the data associated with the generated pack.

Rename the structure to just "generated_pack", resulting in less awkward function names, like "generated_pack_has_ext()" which is preferable to "generated_pack_data_has_ext()".

Rename a few related functions to align with the convention that functions to do with a struct "S" should be prefixed with "S_".

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 32 ++++++++++++++++----------------
 1 file changed, 16 insertions(+), 16 deletions(-)
Show changes to builtin/repack.c +16 −16
diff --git a/builtin/repack.c b/builtin/repack.c
index a168c88791..a4d80b6b04 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -146,15 +146,15 @@ static struct {
 	{".idx"},
 };
 
-struct generated_pack_data {
+struct generated_pack {
 	struct tempfile *tempfiles[ARRAY_SIZE(exts)];
 };
 
-static struct generated_pack_data *populate_pack_exts(const char *name)
+static struct generated_pack *generated_pack_populate(const char *name)
 {
 	struct stat statbuf;
 	struct strbuf path = STRBUF_INIT;
-	struct generated_pack_data *data = xcalloc(1, sizeof(*data));
+	struct generated_pack *pack = xcalloc(1, sizeof(*pack));
 	int i;
 
 	for (i = 0; i < ARRAY_SIZE(exts); i++) {
@@ -164,21 +164,21 @@ static struct generated_pack_data *populate_pack_exts(const char *name)
 		if (stat(path.buf, &statbuf))
 			continue;
 
-		data->tempfiles[i] = register_tempfile(path.buf);
+		pack->tempfiles[i] = register_tempfile(path.buf);
 	}
 
 	strbuf_release(&path);
-	return data;
+	return pack;
 }
 
-static int has_pack_ext(const struct generated_pack_data *data,
-			const char *ext)
+static int generated_pack_has_ext(const struct generated_pack *pack,
+				  const char *ext)
 {
 	int i;
 	for (i = 0; i < ARRAY_SIZE(exts); i++) {
 		if (strcmp(exts[i].name, ext))
 			continue;
-		return !!data->tempfiles[i];
+		return !!pack->tempfiles[i];
 	}
 	BUG("unknown pack extension: '%s'", ext);
 }
@@ -239,7 +239,7 @@ static void repack_promisor_objects(struct repository *repo,
 					  line.buf);
 		write_promisor_file(promisor_name, NULL, 0);
 
-		item->util = populate_pack_exts(item->string);
+		item->util = generated_pack_populate(item->string);
 
 		free(promisor_name);
 	}
@@ -780,8 +780,8 @@ static int write_midx_included_packs(struct string_list *include,
 		 * will suffice, so pick the first one.)
 		 */
 		for_each_string_list_item(item, names) {
-			struct generated_pack_data *data = item->util;
-			if (has_pack_ext(data, ".mtimes"))
+			struct generated_pack *pack = item->util;
+			if (generated_pack_has_ext(pack, ".mtimes"))
 				continue;
 
 			strvec_pushf(&cmd.args, "--preferred-pack=pack-%s.pack",
@@ -864,7 +864,7 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 		 */
 		if (local) {
 			item = string_list_append(names, line.buf);
-			item->util = populate_pack_exts(line.buf);
+			item->util = generated_pack_populate(line.buf);
 		}
 	}
 	fclose(out);
@@ -1435,7 +1435,7 @@ int cmd_repack(int argc,
 	 * Ok we have prepared all new packfiles.
 	 */
 	for_each_string_list_item(item, &names) {
-		struct generated_pack_data *data = item->util;
+		struct generated_pack *pack = item->util;
 
 		for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
 			char *fname;
@@ -1443,8 +1443,8 @@ int cmd_repack(int argc,
 			fname = mkpathdup("%s/pack-%s%s",
 					packdir, item->string, exts[ext].name);
 
-			if (data->tempfiles[ext]) {
-				const char *fname_old = get_tempfile_path(data->tempfiles[ext]);
+			if (pack->tempfiles[ext]) {
+				const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
 				struct stat statbuffer;
 
 				if (!stat(fname_old, &statbuffer)) {
@@ -1452,7 +1452,7 @@ int cmd_repack(int argc,
 					chmod(fname_old, statbuffer.st_mode);
 				}
 
-				if (rename_tempfile(&data->tempfiles[ext], fname))
+				if (rename_tempfile(&pack->tempfiles[ext], fname))
 					die_errno(_("renaming pack to '%s' failed"), fname);
 			} else if (!exts[ext].optional)
 				die(_("pack-objects did not write a '%s' file for pack %s-%s"),
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 21/49] builtin/repack.c: factor our "generated_pack_install"

Once all new packs are known to exist, 'repack' installs their contents from their temporary location into their permanent one. This is a semi-involved procedure for each pack, since for each extension (e.g., ".idx", ".pack", ".mtimes", and so on) we have to either:

 - adjust the filemode of the temporary file before renaming it into
   place, or
 - die() if we are missing a non-optional extension, or
 - unlink() any existing file for extensions that we did not generate
   (e.g., if a non-cruft pack we generated was identical to, say, a
   cruft pack which existed at the beginning of the process, we have to
   remove the ".mtimes" file).

Extract this procedure into its own function, and call it "generated_pack_install"(). This will set us up for pulling this function out of the builtin entirely and making it part of the repack.h API, which will be done in a future commit.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 66 ++++++++++++++++++++++++++----------------------
 1 file changed, 36 insertions(+), 30 deletions(-)
Show changes to builtin/repack.c +36 −30
diff --git a/builtin/repack.c b/builtin/repack.c
index a4d80b6b04..8c3a5f4f80 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -183,6 +183,38 @@ static int generated_pack_has_ext(const struct generated_pack *pack,
 	BUG("unknown pack extension: '%s'", ext);
 }
 
+static void generated_pack_install(struct generated_pack *pack,
+				   const char *name)
+{
+	int ext;
+	for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
+		char *fname;
+
+		fname = mkpathdup("%s/pack-%s%s", packdir, name,
+				  exts[ext].name);
+
+		if (pack->tempfiles[ext]) {
+			const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
+			struct stat statbuffer;
+
+			if (!stat(fname_old, &statbuffer)) {
+				statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
+				chmod(fname_old, statbuffer.st_mode);
+			}
+
+			if (rename_tempfile(&pack->tempfiles[ext], fname))
+				die_errno(_("renaming pack to '%s' failed"),
+					  fname);
+		} else if (!exts[ext].optional)
+			die(_("pack-objects did not write a '%s' file for pack %s-%s"),
+			    exts[ext].name, packtmp, name);
+		else if (unlink(fname) < 0 && errno != ENOENT)
+			die_errno(_("could not unlink: %s"), fname);
+
+		free(fname);
+	}
+}
+
 static void repack_promisor_objects(struct repository *repo,
 				    const struct pack_objects_args *args,
 				    struct string_list *names)
@@ -1045,7 +1077,7 @@ int cmd_repack(int argc,
 	struct existing_packs existing = EXISTING_PACKS_INIT;
 	struct pack_geometry geometry = { 0 };
 	struct tempfile *refs_snapshot = NULL;
-	int i, ext, ret;
+	int i, ret;
 	int show_progress;
 	char **midx_pack_names = NULL;
 	size_t midx_pack_names_nr = 0;
@@ -1434,35 +1466,9 @@ int cmd_repack(int argc,
 	/*
 	 * Ok we have prepared all new packfiles.
 	 */
-	for_each_string_list_item(item, &names) {
-		struct generated_pack *pack = item->util;
-
-		for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
-			char *fname;
-
-			fname = mkpathdup("%s/pack-%s%s",
-					packdir, item->string, exts[ext].name);
-
-			if (pack->tempfiles[ext]) {
-				const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
-				struct stat statbuffer;
-
-				if (!stat(fname_old, &statbuffer)) {
-					statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
-					chmod(fname_old, statbuffer.st_mode);
-				}
-
-				if (rename_tempfile(&pack->tempfiles[ext], fname))
-					die_errno(_("renaming pack to '%s' failed"), fname);
-			} else if (!exts[ext].optional)
-				die(_("pack-objects did not write a '%s' file for pack %s-%s"),
-				    exts[ext].name, packtmp, item->string);
-			else if (unlink(fname) < 0 && errno != ENOENT)
-				die_errno(_("could not unlink: %s"), fname);
-
-			free(fname);
-		}
-	}
+	for_each_string_list_item(item, &names)
+		generated_pack_install((struct generated_pack *)item->util,
+				       item->string);
 	/* End of pack replacement. */
 
 	if (delete_redundant && pack_everything & ALL_INTO_ONE)
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Sep 29, 2025, 23:21 UTC · re: Taylor Blau · lore

Re: [PATCH 21/49] builtin/repack.c: factor our "generated_pack_install"

On Sun, Sep 28, 2025 at 06:08:31PM -0400, Taylor Blau wrote:
Is the "factor our" in the commit subject supposed to read "factor out"?
Show 39 quoted lines
> diff --git a/builtin/repack.c b/builtin/repack.c
> index a4d80b6b04..8c3a5f4f80 100644
> --- a/builtin/repack.c
> +++ b/builtin/repack.c
> @@ -1434,35 +1466,9 @@ int cmd_repack(int argc,
>  	/*
>  	 * Ok we have prepared all new packfiles.
>  	 */
> -	for_each_string_list_item(item, &names) {
> -		struct generated_pack *pack = item->util;
> -
> -		for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
> -			char *fname;
> -
> -			fname = mkpathdup("%s/pack-%s%s",
> -					packdir, item->string, exts[ext].name);
> -
> -			if (pack->tempfiles[ext]) {
> -				const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
> -				struct stat statbuffer;
> -
> -				if (!stat(fname_old, &statbuffer)) {
> -					statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
> -					chmod(fname_old, statbuffer.st_mode);
> -				}
> -
> -				if (rename_tempfile(&pack->tempfiles[ext], fname))
> -					die_errno(_("renaming pack to '%s' failed"), fname);
> -			} else if (!exts[ext].optional)
> -				die(_("pack-objects did not write a '%s' file for pack %s-%s"),
> -				    exts[ext].name, packtmp, item->string);
> -			else if (unlink(fname) < 0 && errno != ENOENT)
> -				die_errno(_("could not unlink: %s"), fname);
> -
> -			free(fname);
> -		}
> -	}
> +	for_each_string_list_item(item, &names)
> +		generated_pack_install((struct generated_pack *)item->util,

This cast should be unnecessary, right? `item->util` is a void pointer, so C should do the cast implicitly.

Patrick
Taylor Blau· Oct 7, 2025, 20:26 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 21/49] builtin/repack.c: factor our "generated_pack_install"

On Tue, Sep 30, 2025 at 01:21:46AM +0200, Patrick Steinhardt wrote:
> On Sun, Sep 28, 2025 at 06:08:31PM -0400, Taylor Blau wrote:
>
> Is the "factor our" in the commit subject supposed to read "factor out"?
Oops, good catch ;-).
Show 5 quoted lines
> > +	for_each_string_list_item(item, &names)
> > +		generated_pack_install((struct generated_pack *)item->util,
>
> This cast should be unnecessary, right? `item->util` is a void pointer,
> so C should do the cast implicitly.

It's unnecessary, but I dislike implicit casts from 'void*' to any other type. This makes it clearer how we're supposed to interpret the value in item->util, but I'm happy to change it to use the implicit cast if you feel strongly about it.

Thanks, Taylor

Patrick Steinhardt· Oct 8, 2025, 04:28 UTC · re: Taylor Blau · lore

Re: [PATCH 21/49] builtin/repack.c: factor our "generated_pack_install"

On Tue, Oct 07, 2025 at 04:26:26PM -0400, Taylor Blau wrote:
Show 12 quoted lines
> On Tue, Sep 30, 2025 at 01:21:46AM +0200, Patrick Steinhardt wrote:
> > On Sun, Sep 28, 2025 at 06:08:31PM -0400, Taylor Blau wrote:
> > > +	for_each_string_list_item(item, &names)
> > > +		generated_pack_install((struct generated_pack *)item->util,
> >
> > This cast should be unnecessary, right? `item->util` is a void pointer,
> > so C should do the cast implicitly.
> 
> It's unnecessary, but I dislike implicit casts from 'void*' to any other
> type. This makes it clearer how we're supposed to interpret the value in
> item->util, but I'm happy to change it to use the implicit cast if you
> feel strongly about it.

I think it's common practice to not have such an explicit cast in our code base. But in any case, I don't feel strongly about this.

Patrick
Jeff King· Oct 10, 2025, 06:14 UTC · re: Taylor Blau · lore

Re: [PATCH 21/49] builtin/repack.c: factor our "generated_pack_install"

On Tue, Oct 07, 2025 at 04:26:26PM -0400, Taylor Blau wrote:
Show 10 quoted lines
> > > +	for_each_string_list_item(item, &names)
> > > +		generated_pack_install((struct generated_pack *)item->util,
> >
> > This cast should be unnecessary, right? `item->util` is a void pointer,
> > so C should do the cast implicitly.
> 
> It's unnecessary, but I dislike implicit casts from 'void*' to any other
> type. This makes it clearer how we're supposed to interpret the value in
> item->util, but I'm happy to change it to use the implicit cast if you
> feel strongly about it.

I tend to avoid casts when possible, because they can mask unexpected conversions. E.g., if item->util's type changed, your cast means we would never notice it.

But that cuts both ways. If the prototype for generated_pack_install() changes, it is only your cast that would tell the compiler that this caller needs to be updated. Ultimately it is the spot that _sets_ item->util that needs to change, but hopefully flagging this spot would point us in the right direction.

And that does seem more likely than item->util changing away from void. (To be clear, I don't think either is that likely, I'm just musing on whether casts like this are helpful in general).

-Peff
Taylor Blau· Oct 10, 2025, 22:58 UTC · re: Jeff King · lore

Re: [PATCH 21/49] builtin/repack.c: factor our "generated_pack_install"

On Fri, Oct 10, 2025 at 02:14:49AM -0400, Jeff King wrote:
Show 26 quoted lines
> On Tue, Oct 07, 2025 at 04:26:26PM -0400, Taylor Blau wrote:
>
> > > > +	for_each_string_list_item(item, &names)
> > > > +		generated_pack_install((struct generated_pack *)item->util,
> > >
> > > This cast should be unnecessary, right? `item->util` is a void pointer,
> > > so C should do the cast implicitly.
> >
> > It's unnecessary, but I dislike implicit casts from 'void*' to any other
> > type. This makes it clearer how we're supposed to interpret the value in
> > item->util, but I'm happy to change it to use the implicit cast if you
> > feel strongly about it.
>
> I tend to avoid casts when possible, because they can mask unexpected
> conversions. E.g., if item->util's type changed, your cast means we
> would never notice it.
>
> But that cuts both ways. If the prototype for generated_pack_install()
> changes, it is only your cast that would tell the compiler that this
> caller needs to be updated. Ultimately it is the spot that _sets_
> item->util that needs to change, but hopefully flagging this spot would
> point us in the right direction.
>
> And that does seem more likely than item->util changing away from void.
> (To be clear, I don't think either is that likely, I'm just musing on
> whether casts like this are helpful in general).

OK, that's a very compelling argument. I'll drop the cast on my local version.

(As a side note, it doesn't look like we mention this convention in our CodingGuidelines, and we probably should, including the rationale that you gave above here. That seems like another good candidate for #leftoverbits.)

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 22/49] builtin/repack.c: pass "packtmp" to `generated_pack_populate()`

In a similar spirit as previous commits, this function needs to know the temporary pack prefix, which it currently accesses through the static "packtmp" variable within builtin/repack.c.

Pass it explicitly as a function parameter to facilitate moving this function out of builtin/repack.c entirely.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
Show changes to builtin/repack.c +4 −3
diff --git a/builtin/repack.c b/builtin/repack.c
index 8c3a5f4f80..2141c43bd2 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -150,7 +150,8 @@ struct generated_pack {
 	struct tempfile *tempfiles[ARRAY_SIZE(exts)];
 };
 
-static struct generated_pack *generated_pack_populate(const char *name)
+static struct generated_pack *generated_pack_populate(const char *name,
+						      const char *packtmp)
 {
 	struct stat statbuf;
 	struct strbuf path = STRBUF_INIT;
@@ -271,7 +272,7 @@ static void repack_promisor_objects(struct repository *repo,
 					  line.buf);
 		write_promisor_file(promisor_name, NULL, 0);
 
-		item->util = generated_pack_populate(item->string);
+		item->util = generated_pack_populate(item->string, packtmp);
 
 		free(promisor_name);
 	}
@@ -896,7 +897,7 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 		 */
 		if (local) {
 			item = string_list_append(names, line.buf);
-			item->util = generated_pack_populate(line.buf);
+			item->util = generated_pack_populate(line.buf, packtmp);
 		}
 	}
 	fclose(out);
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Sep 29, 2025, 23:21 UTC · re: Taylor Blau · lore

Re: [PATCH 22/49] builtin/repack.c: pass "packtmp" to `generated_pack_populate()`

On Sun, Sep 28, 2025 at 06:08:35PM -0400, Taylor Blau wrote:
> In a similar spirit as previous commits, this function needs to know the
> temporary pack prefix, which it currently accesses through the static
s/accesses/accessed/
Patrick
Taylor Blau· Oct 7, 2025, 20:29 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 22/49] builtin/repack.c: pass "packtmp" to `generated_pack_populate()`

On Tue, Sep 30, 2025 at 01:21:53AM +0200, Patrick Steinhardt wrote:
Show 5 quoted lines
> On Sun, Sep 28, 2025 at 06:08:35PM -0400, Taylor Blau wrote:
> > In a similar spirit as previous commits, this function needs to know the
> > temporary pack prefix, which it currently accesses through the static
>
> s/accesses/accessed/

I think that the original is correct here. "it currently accesses" as opposed to "it currently accessed". I have read it back a couple of times, but it is entirely possible that I'm missing something ;-). LMK.

Thanks, Taylor

Patrick Steinhardt· Oct 8, 2025, 04:27 UTC · re: Taylor Blau · lore

Re: [PATCH 22/49] builtin/repack.c: pass "packtmp" to `generated_pack_populate()`

On Tue, Oct 07, 2025 at 04:29:52PM -0400, Taylor Blau wrote:
Show 10 quoted lines
> On Tue, Sep 30, 2025 at 01:21:53AM +0200, Patrick Steinhardt wrote:
> > On Sun, Sep 28, 2025 at 06:08:35PM -0400, Taylor Blau wrote:
> > > In a similar spirit as previous commits, this function needs to know the
> > > temporary pack prefix, which it currently accesses through the static
> >
> > s/accesses/accessed/
> 
> I think that the original is correct here. "it currently accesses" as
> opposed to "it currently accessed". I have read it back a couple of
> times, but it is entirely possible that I'm missing something ;-). LMK.
Huh, yeah, no idea what I was thinking here.
Patrick
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 23/49] builtin/repack.c: provide pack locations to `generated_pack_install()`

Repeat what was done in the preceding commit for the `generated_pack_install()` function, which needs both "packdir" and "packtmp".

(As an aside, it is somewhat unfortunate that the final three parameters to this function are all "const char *", making errors like passing "packdir" and "packtmp" in the wrong order easy. We could define a new structure here, but that may be too heavy-handed.)

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +3 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index 2141c43bd2..a4f0a19453 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -185,7 +185,8 @@ static int generated_pack_has_ext(const struct generated_pack *pack,
 }
 
 static void generated_pack_install(struct generated_pack *pack,
-				   const char *name)
+				   const char *name,
+				   const char *packdir, const char *packtmp)
 {
 	int ext;
 	for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
@@ -1469,7 +1470,7 @@ int cmd_repack(int argc,
 	 */
 	for_each_string_list_item(item, &names)
 		generated_pack_install((struct generated_pack *)item->util,
-				       item->string);
+				       item->string, packdir, packtmp);
 	/* End of pack replacement. */
 
 	if (delete_redundant && pack_everything & ALL_INTO_ONE)
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 24/49] repack: remove 'generated_pack' API from the builtin

Now that we have factored the "generated_pack" API, we can move it to repack.ch, further slimming down builtin/repack.c.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 83 ------------------------------------------------
 repack.c         | 83 ++++++++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  8 +++++
 3 files changed, 91 insertions(+), 83 deletions(-)
Show changes to 3 files +91 −83

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index a4f0a19453..b7826e676b 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -134,89 +134,6 @@ static int write_oid(const struct object_id *oid,
 	return 0;
 }
 
-static struct {
-	const char *name;
-	unsigned optional:1;
-} exts[] = {
-	{".pack"},
-	{".rev", 1},
-	{".mtimes", 1},
-	{".bitmap", 1},
-	{".promisor", 1},
-	{".idx"},
-};
-
-struct generated_pack {
-	struct tempfile *tempfiles[ARRAY_SIZE(exts)];
-};
-
-static struct generated_pack *generated_pack_populate(const char *name,
-						      const char *packtmp)
-{
-	struct stat statbuf;
-	struct strbuf path = STRBUF_INIT;
-	struct generated_pack *pack = xcalloc(1, sizeof(*pack));
-	int i;
-
-	for (i = 0; i < ARRAY_SIZE(exts); i++) {
-		strbuf_reset(&path);
-		strbuf_addf(&path, "%s-%s%s", packtmp, name, exts[i].name);
-
-		if (stat(path.buf, &statbuf))
-			continue;
-
-		pack->tempfiles[i] = register_tempfile(path.buf);
-	}
-
-	strbuf_release(&path);
-	return pack;
-}
-
-static int generated_pack_has_ext(const struct generated_pack *pack,
-				  const char *ext)
-{
-	int i;
-	for (i = 0; i < ARRAY_SIZE(exts); i++) {
-		if (strcmp(exts[i].name, ext))
-			continue;
-		return !!pack->tempfiles[i];
-	}
-	BUG("unknown pack extension: '%s'", ext);
-}
-
-static void generated_pack_install(struct generated_pack *pack,
-				   const char *name,
-				   const char *packdir, const char *packtmp)
-{
-	int ext;
-	for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
-		char *fname;
-
-		fname = mkpathdup("%s/pack-%s%s", packdir, name,
-				  exts[ext].name);
-
-		if (pack->tempfiles[ext]) {
-			const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
-			struct stat statbuffer;
-
-			if (!stat(fname_old, &statbuffer)) {
-				statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
-				chmod(fname_old, statbuffer.st_mode);
-			}
-
-			if (rename_tempfile(&pack->tempfiles[ext], fname))
-				die_errno(_("renaming pack to '%s' failed"),
-					  fname);
-		} else if (!exts[ext].optional)
-			die(_("pack-objects did not write a '%s' file for pack %s-%s"),
-			    exts[ext].name, packtmp, name);
-		else if (unlink(fname) < 0 && errno != ENOENT)
-			die_errno(_("could not unlink: %s"), fname);
-
-		free(fname);
-	}
-}
-
 static void repack_promisor_objects(struct repository *repo,
 				    const struct pack_objects_args *args,
 				    struct string_list *names)
diff --git a/repack.c b/repack.c
index 9182e1c50b..d8afdd352d 100644
--- a/repack.c
+++ b/repack.c
@@ -3,9 +3,11 @@
 #include "midx.h"
 #include "odb.h"
 #include "packfile.h"
+#include "path.h"
 #include "repack.h"
 #include "repository.h"
 #include "run-command.h"
+#include "tempfile.h"
 
 void prepare_pack_objects(struct child_process *cmd,
 			  const struct pack_objects_args *args,
@@ -219,3 +221,84 @@ void existing_packs_release(struct existing_packs *existing)
 	string_list_clear(&existing->non_kept_packs, 0);
 	string_list_clear(&existing->cruft_packs, 0);
 }
+
+static struct {
+	const char *name;
+	unsigned optional:1;
+} exts[] = {
+	{".pack"},
+	{".rev", 1},
+	{".mtimes", 1},
+	{".bitmap", 1},
+	{".promisor", 1},
+	{".idx"},
+};
+
+struct generated_pack {
+	struct tempfile *tempfiles[ARRAY_SIZE(exts)];
+};
+
+struct generated_pack *generated_pack_populate(const char *name,
+					       const char *packtmp)
+{
+	struct stat statbuf;
+	struct strbuf path = STRBUF_INIT;
+	struct generated_pack *pack = xcalloc(1, sizeof(*pack));
+	size_t i;
+
+	for (i = 0; i < ARRAY_SIZE(exts); i++) {
+		strbuf_reset(&path);
+		strbuf_addf(&path, "%s-%s%s", packtmp, name, exts[i].name);
+
+		if (stat(path.buf, &statbuf))
+			continue;
+
+		pack->tempfiles[i] = register_tempfile(path.buf);
+	}
+
+	strbuf_release(&path);
+	return pack;
+}
+
+int generated_pack_has_ext(const struct generated_pack *pack, const char *ext)
+{
+	size_t i;
+	for (i = 0; i < ARRAY_SIZE(exts); i++) {
+		if (strcmp(exts[i].name, ext))
+			continue;
+		return !!pack->tempfiles[i];
+	}
+	BUG("unknown pack extension: '%s'", ext);
+}
+
+void generated_pack_install(struct generated_pack *pack, const char *name,
+			    const char *packdir, const char *packtmp)
+{
+	size_t ext;
+	for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
+		char *fname;
+
+		fname = mkpathdup("%s/pack-%s%s", packdir, name,
+				  exts[ext].name);
+
+		if (pack->tempfiles[ext]) {
+			const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
+			struct stat statbuffer;
+
+			if (!stat(fname_old, &statbuffer)) {
+				statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
+				chmod(fname_old, statbuffer.st_mode);
+			}
+
+			if (rename_tempfile(&pack->tempfiles[ext], fname))
+				die_errno(_("renaming pack to '%s' failed"),
+					  fname);
+		} else if (!exts[ext].optional)
+			die(_("pack-objects did not write a '%s' file for pack %s-%s"),
+			    exts[ext].name, packtmp, name);
+		else if (unlink(fname) < 0 && errno != ENOENT)
+			die_errno(_("could not unlink: %s"), fname);
+
+		free(fname);
+	}
+}
diff --git a/repack.h b/repack.h
index 19796e2243..f37eb49524 100644
--- a/repack.h
+++ b/repack.h
@@ -66,4 +66,12 @@ void existing_packs_remove_redundant(struct existing_packs *existing,
 				     const char *packdir);
 void existing_packs_release(struct existing_packs *existing);
 
+struct generated_pack;
+
+struct generated_pack *generated_pack_populate(const char *name,
+					       const char *packtmp);
+int generated_pack_has_ext(const struct generated_pack *pack, const char *ext);
+void generated_pack_install(struct generated_pack *pack, const char *name,
+			    const char *packdir, const char *packtmp);
+
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 25/49] builtin/repack.c: pass "packtmp" to `repack_promisor_objects()`

In a similar spirit as previous commit(s), pass the "packtmp" variable to "repack_promisor_objects()" as an explicit parameter of the function, preparing us to move this function in a following commit.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +3 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index b7826e676b..ca7658e38f 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -136,7 +136,8 @@ static int write_oid(const struct object_id *oid,
 
 static void repack_promisor_objects(struct repository *repo,
 				    const struct pack_objects_args *args,
-				    struct string_list *names)
+				    struct string_list *names,
+				    const char *packtmp)
 {
 	struct write_oid_context ctx;
 	struct child_process cmd = CHILD_PROCESS_INIT;
@@ -1199,7 +1200,7 @@ int cmd_repack(int argc,
 		strvec_push(&cmd.args, "--delta-islands");
 
 	if (pack_everything & ALL_INTO_ONE) {
-		repack_promisor_objects(repo, &po_args, &names);
+		repack_promisor_objects(repo, &po_args, &names, packtmp);
 
 		if (existing_packs_has_non_kept(&existing) &&
 		    delete_redundant &&
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 26/49] builtin/repack.c: remove "repack_promisor_objects()" from the builtin

Now that we have properly factored the portion of the builtin which is responsible for repacking promisor objects, we can move that function (and associated dependencies) out of the builtin entirely.

Similar to previous extractions, this function is declared in repack.h, but implemented in a separate repack-promisor.c file. This is done to separate promisor-specific repacking functionality from generic repack utilities (like "existing_packs", and "generated_pack" APIs).

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile          |   1 +
 builtin/repack.c  |  95 ------------------------------------------
 meson.build       |   1 +
 repack-promisor.c | 102 ++++++++++++++++++++++++++++++++++++++++++++++
 repack.h          |   4 ++
 5 files changed, 108 insertions(+), 95 deletions(-)
 create mode 100644 repack-promisor.c
Show changes to 5 files +108 −95

Makefile, builtin/repack.c, meson.build, repack-promisor.c, repack.h

diff --git a/Makefile b/Makefile
index b220f0f7d8..0a73f8adaf 100644
--- a/Makefile
+++ b/Makefile
@@ -1137,6 +1137,7 @@ LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
+LIB_OBJS += repack-promisor.o
 LIB_OBJS += replace-object.o
 LIB_OBJS += repo-settings.o
 LIB_OBJS += repository.o
diff --git a/builtin/repack.c b/builtin/repack.c
index ca7658e38f..18c3df7200 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,101 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-struct write_oid_context {
-	struct child_process *cmd;
-	const struct git_hash_algo *algop;
-};
-
-/*
- * Write oid to the given struct child_process's stdin, starting it first if
- * necessary.
- */
-static int write_oid(const struct object_id *oid,
-		     struct packed_git *pack UNUSED,
-		     uint32_t pos UNUSED, void *data)
-{
-	struct write_oid_context *ctx = data;
-	struct child_process *cmd = ctx->cmd;
-
-	if (cmd->in == -1) {
-		if (start_command(cmd))
-			die(_("could not start pack-objects to repack promisor objects"));
-	}
-
-	if (write_in_full(cmd->in, oid_to_hex(oid), ctx->algop->hexsz) < 0 ||
-	    write_in_full(cmd->in, "\n", 1) < 0)
-		die(_("failed to feed promisor objects to pack-objects"));
-	return 0;
-}
-
-static void repack_promisor_objects(struct repository *repo,
-				    const struct pack_objects_args *args,
-				    struct string_list *names,
-				    const char *packtmp)
-{
-	struct write_oid_context ctx;
-	struct child_process cmd = CHILD_PROCESS_INIT;
-	FILE *out;
-	struct strbuf line = STRBUF_INIT;
-
-	prepare_pack_objects(&cmd, args, packtmp);
-	cmd.in = -1;
-
-	/*
-	 * NEEDSWORK: Giving pack-objects only the OIDs without any ordering
-	 * hints may result in suboptimal deltas in the resulting pack. See if
-	 * the OIDs can be sent with fake paths such that pack-objects can use a
-	 * {type -> existing pack order} ordering when computing deltas instead
-	 * of a {type -> size} ordering, which may produce better deltas.
-	 */
-	ctx.cmd = &cmd;
-	ctx.algop = repo->hash_algo;
-	for_each_packed_object(repo, write_oid, &ctx,
-			       FOR_EACH_OBJECT_PROMISOR_ONLY);
-
-	if (cmd.in == -1) {
-		/* No packed objects; cmd was never started */
-		child_process_clear(&cmd);
-		return;
-	}
-
-	close(cmd.in);
-
-	out = xfdopen(cmd.out, "r");
-	while (strbuf_getline_lf(&line, out) != EOF) {
-		struct string_list_item *item;
-		char *promisor_name;
-
-		if (line.len != repo->hash_algo->hexsz)
-			die(_("repack: Expecting full hex object ID lines only from pack-objects."));
-		item = string_list_append(names, line.buf);
-
-		/*
-		 * pack-objects creates the .pack and .idx files, but not the
-		 * .promisor file. Create the .promisor file, which is empty.
-		 *
-		 * NEEDSWORK: fetch-pack sometimes generates non-empty
-		 * .promisor files containing the ref names and associated
-		 * hashes at the point of generation of the corresponding
-		 * packfile, but this would not preserve their contents. Maybe
-		 * concatenate the contents of all .promisor files instead of
-		 * just creating a new empty file.
-		 */
-		promisor_name = mkpathdup("%s-%s.promisor", packtmp,
-					  line.buf);
-		write_promisor_file(promisor_name, NULL, 0);
-
-		item->util = generated_pack_populate(item->string, packtmp);
-
-		free(promisor_name);
-	}
-
-	fclose(out);
-	if (finish_command(&cmd))
-		die(_("could not finish pack-objects to repack promisor objects"));
-	strbuf_release(&line);
-}
-
 struct pack_geometry {
 	struct packed_git **pack;
 	uint32_t pack_nr, pack_alloc;
diff --git a/meson.build b/meson.build
index 8ab287a2ed..ca5603b456 100644
--- a/meson.build
+++ b/meson.build
@@ -463,6 +463,7 @@ libgit_sources = [
   'reftable/writer.c',
   'remote.c',
   'repack.c',
+  'repack-promisor.c',
   'replace-object.c',
   'repo-settings.c',
   'repository.c',
diff --git a/repack-promisor.c b/repack-promisor.c
new file mode 100644
index 0000000000..8bf42fc715
--- /dev/null
+++ b/repack-promisor.c
@@ -0,0 +1,102 @@
+#include "git-compat-util.h"
+#include "repack.h"
+#include "run-command.h"
+#include "hex.h"
+#include "repository.h"
+#include "packfile.h"
+#include "path.h"
+#include "pack.h"
+
+struct write_oid_context {
+	struct child_process *cmd;
+	const struct git_hash_algo *algop;
+};
+
+/*
+ * Write oid to the given struct child_process's stdin, starting it first if
+ * necessary.
+ */
+static int write_oid(const struct object_id *oid,
+		     struct packed_git *pack UNUSED,
+		     uint32_t pos UNUSED, void *data)
+{
+	struct write_oid_context *ctx = data;
+	struct child_process *cmd = ctx->cmd;
+
+	if (cmd->in == -1) {
+		if (start_command(cmd))
+			die(_("could not start pack-objects to repack promisor objects"));
+	}
+
+	if (write_in_full(cmd->in, oid_to_hex(oid), ctx->algop->hexsz) < 0 ||
+	    write_in_full(cmd->in, "\n", 1) < 0)
+		die(_("failed to feed promisor objects to pack-objects"));
+	return 0;
+}
+
+void repack_promisor_objects(struct repository *repo,
+			     const struct pack_objects_args *args,
+			     struct string_list *names, const char *packtmp)
+{
+	struct write_oid_context ctx;
+	struct child_process cmd = CHILD_PROCESS_INIT;
+	FILE *out;
+	struct strbuf line = STRBUF_INIT;
+
+	prepare_pack_objects(&cmd, args, packtmp);
+	cmd.in = -1;
+
+	/*
+	 * NEEDSWORK: Giving pack-objects only the OIDs without any ordering
+	 * hints may result in suboptimal deltas in the resulting pack. See if
+	 * the OIDs can be sent with fake paths such that pack-objects can use a
+	 * {type -> existing pack order} ordering when computing deltas instead
+	 * of a {type -> size} ordering, which may produce better deltas.
+	 */
+	ctx.cmd = &cmd;
+	ctx.algop = repo->hash_algo;
+	for_each_packed_object(repo, write_oid, &ctx,
+			       FOR_EACH_OBJECT_PROMISOR_ONLY);
+
+	if (cmd.in == -1) {
+		/* No packed objects; cmd was never started */
+		child_process_clear(&cmd);
+		return;
+	}
+
+	close(cmd.in);
+
+	out = xfdopen(cmd.out, "r");
+	while (strbuf_getline_lf(&line, out) != EOF) {
+		struct string_list_item *item;
+		char *promisor_name;
+
+		if (line.len != repo->hash_algo->hexsz)
+			die(_("repack: Expecting full hex object ID lines only from pack-objects."));
+		item = string_list_append(names, line.buf);
+
+		/*
+		 * pack-objects creates the .pack and .idx files, but not the
+		 * .promisor file. Create the .promisor file, which is empty.
+		 *
+		 * NEEDSWORK: fetch-pack sometimes generates non-empty
+		 * .promisor files containing the ref names and associated
+		 * hashes at the point of generation of the corresponding
+		 * packfile, but this would not preserve their contents. Maybe
+		 * concatenate the contents of all .promisor files instead of
+		 * just creating a new empty file.
+		 */
+		promisor_name = mkpathdup("%s-%s.promisor", packtmp,
+					  line.buf);
+		write_promisor_file(promisor_name, NULL, 0);
+
+		item->util = generated_pack_populate(item->string, packtmp);
+
+		free(promisor_name);
+	}
+
+	fclose(out);
+	if (finish_command(&cmd))
+		die(_("could not finish pack-objects to repack promisor objects"));
+	strbuf_release(&line);
+}
diff --git a/repack.h b/repack.h
index f37eb49524..19dc4fd738 100644
--- a/repack.h
+++ b/repack.h
@@ -74,4 +74,8 @@ int generated_pack_has_ext(const struct generated_pack *pack, const char *ext);
 void generated_pack_install(struct generated_pack *pack, const char *name,
 			    const char *packdir, const char *packtmp);
 
+void repack_promisor_objects(struct repository *repo,
+			     const struct pack_objects_args *args,
+			     struct string_list *names, const char *packtmp);
+
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Sep 29, 2025, 23:21 UTC · re: Taylor Blau · lore

Re: [PATCH 26/49] builtin/repack.c: remove "repack_promisor_objects()" from the builtin

On Sun, Sep 28, 2025 at 06:08:51PM -0400, Taylor Blau wrote:
Show 8 quoted lines
> Now that we have properly factored the portion of the builtin which is
> responsible for repacking promisor objects, we can move that function
> (and associated dependencies) out of the builtin entirely.
> 
> Similar to previous extractions, this function is declared in repack.h,
> but implemented in a separate repack-promisor.c file. This is done to
> separate promisor-specific repacking functionality from generic repack
> utilities (like "existing_packs", and "generated_pack" APIs).
Feels sensible.
Show 14 quoted lines
> diff --git a/repack-promisor.c b/repack-promisor.c
> new file mode 100644
> index 0000000000..8bf42fc715
> --- /dev/null
> +++ b/repack-promisor.c
> @@ -0,0 +1,102 @@
> +#include "git-compat-util.h"
> +#include "repack.h"
> +#include "run-command.h"
> +#include "hex.h"
> +#include "repository.h"
> +#include "packfile.h"
> +#include "path.h"
> +#include "pack.h"
Nit: these headers should probably be sorted lexicographically?
Show 13 quoted lines
> diff --git a/repack.h b/repack.h
> index f37eb49524..19dc4fd738 100644
> --- a/repack.h
> +++ b/repack.h
> @@ -74,4 +74,8 @@ int generated_pack_has_ext(const struct generated_pack *pack, const char *ext);
>  void generated_pack_install(struct generated_pack *pack, const char *name,
>  			    const char *packdir, const char *packtmp);
>  
> +void repack_promisor_objects(struct repository *repo,
> +			     const struct pack_objects_args *args,
> +			     struct string_list *names, const char *packtmp);
> +
>  #endif /* REPACK_H */

I'm not a huge fan of the code bein gin "repack-promisor.c" while the declarations are in "repack.h", as it makes things harder to find from my point of view. But I don't feel too strongly about this.

Patrick
Taylor Blau· Oct 7, 2025, 20:36 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 26/49] builtin/repack.c: remove "repack_promisor_objects()" from the builtin

On Tue, Sep 30, 2025 at 01:21:59AM +0200, Patrick Steinhardt wrote:
Show 16 quoted lines
> > diff --git a/repack-promisor.c b/repack-promisor.c
> > new file mode 100644
> > index 0000000000..8bf42fc715
> > --- /dev/null
> > +++ b/repack-promisor.c
> > @@ -0,0 +1,102 @@
> > +#include "git-compat-util.h"
> > +#include "repack.h"
> > +#include "run-command.h"
> > +#include "hex.h"
> > +#include "repository.h"
> > +#include "packfile.h"
> > +#include "path.h"
> > +#include "pack.h"
>
> Nit: these headers should probably be sorted lexicographically?
Yup, definitely. Fixed.
Show 17 quoted lines
> > diff --git a/repack.h b/repack.h
> > index f37eb49524..19dc4fd738 100644
> > --- a/repack.h
> > +++ b/repack.h
> > @@ -74,4 +74,8 @@ int generated_pack_has_ext(const struct generated_pack *pack, const char *ext);
> >  void generated_pack_install(struct generated_pack *pack, const char *name,
> >  			    const char *packdir, const char *packtmp);
> >
> > +void repack_promisor_objects(struct repository *repo,
> > +			     const struct pack_objects_args *args,
> > +			     struct string_list *names, const char *packtmp);
> > +
> >  #endif /* REPACK_H */
>
> I'm not a huge fan of the code bein gin "repack-promisor.c" while the
> declarations are in "repack.h", as it makes things harder to find from
> my point of view. But I don't feel too strongly about this.

I think we have precedence for this in other places throughout the code (for e.g., "midx.h" => "midx.c" and "midx-write.c", "pack-bitmap.h" => "pack-bitmap.c" and "pack-bitmap-write.c", etc), so I think it's OK here.

If you change your mind and do feel strongly about it later on, I'm happy to adjust it ;-).

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 27/49] builtin/repack.c: rename various pack_geometry functions

Rename functions which work with 'struct pack_geometry' to begin with "pack_geometry_".

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 52 ++++++++++++++++++++++++------------------------
 1 file changed, 26 insertions(+), 26 deletions(-)
Show changes to builtin/repack.c +26 −26
diff --git a/builtin/repack.c b/builtin/repack.c
index 18c3df7200..2ce1ae3364 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -115,17 +115,17 @@ struct pack_geometry {
 	int split_factor;
 };
 
-static uint32_t geometry_pack_weight(struct packed_git *p)
+static uint32_t pack_geometry_weight(struct packed_git *p)
 {
 	if (open_pack_index(p))
 		die(_("cannot open index for %s"), p->pack_name);
 	return p->num_objects;
 }
 
-static int geometry_cmp(const void *va, const void *vb)
+static int pack_geometry_cmp(const void *va, const void *vb)
 {
-	uint32_t aw = geometry_pack_weight(*(struct packed_git **)va),
-		 bw = geometry_pack_weight(*(struct packed_git **)vb);
+	uint32_t aw = pack_geometry_weight(*(struct packed_git **)va),
+		 bw = pack_geometry_weight(*(struct packed_git **)vb);
 
 	if (aw < bw)
 		return -1;
@@ -134,7 +134,7 @@ static int geometry_cmp(const void *va, const void *vb)
 	return 0;
 }
 
-static void init_pack_geometry(struct pack_geometry *geometry,
+static void pack_geometry_init(struct pack_geometry *geometry,
 			       struct existing_packs *existing,
 			       const struct pack_objects_args *args)
 {
@@ -184,11 +184,11 @@ static void init_pack_geometry(struct pack_geometry *geometry,
 		geometry->pack_nr++;
 	}
 
-	QSORT(geometry->pack, geometry->pack_nr, geometry_cmp);
+	QSORT(geometry->pack, geometry->pack_nr, pack_geometry_cmp);
 	strbuf_release(&buf);
 }
 
-static void split_pack_geometry(struct pack_geometry *geometry)
+static void pack_geometry_split(struct pack_geometry *geometry)
 {
 	uint32_t i;
 	uint32_t split;
@@ -208,13 +208,13 @@ static void split_pack_geometry(struct pack_geometry *geometry)
 		struct packed_git *prev = geometry->pack[i - 1];
 
 		if (unsigned_mult_overflows(geometry->split_factor,
-					    geometry_pack_weight(prev)))
+					    pack_geometry_weight(prev)))
 			die(_("pack %s too large to consider in geometric "
 			      "progression"),
 			    prev->pack_name);
 
-		if (geometry_pack_weight(ours) <
-		    geometry->split_factor * geometry_pack_weight(prev))
+		if (pack_geometry_weight(ours) <
+		    geometry->split_factor * pack_geometry_weight(prev))
 			break;
 	}
 
@@ -242,9 +242,9 @@ static void split_pack_geometry(struct pack_geometry *geometry)
 	for (i = 0; i < split; i++) {
 		struct packed_git *p = geometry->pack[i];
 
-		if (unsigned_add_overflows(total_size, geometry_pack_weight(p)))
+		if (unsigned_add_overflows(total_size, pack_geometry_weight(p)))
 			die(_("pack %s too large to roll up"), p->pack_name);
-		total_size += geometry_pack_weight(p);
+		total_size += pack_geometry_weight(p);
 	}
 	for (i = split; i < geometry->pack_nr; i++) {
 		struct packed_git *ours = geometry->pack[i];
@@ -253,15 +253,15 @@ static void split_pack_geometry(struct pack_geometry *geometry)
 					    total_size))
 			die(_("pack %s too large to roll up"), ours->pack_name);
 
-		if (geometry_pack_weight(ours) <
+		if (pack_geometry_weight(ours) <
 		    geometry->split_factor * total_size) {
 			if (unsigned_add_overflows(total_size,
-						   geometry_pack_weight(ours)))
+						   pack_geometry_weight(ours)))
 				die(_("pack %s too large to roll up"),
 				    ours->pack_name);
 
 			split++;
-			total_size += geometry_pack_weight(ours);
+			total_size += pack_geometry_weight(ours);
 		} else
 			break;
 	}
@@ -269,7 +269,7 @@ static void split_pack_geometry(struct pack_geometry *geometry)
 	geometry->split = split;
 }
 
-static struct packed_git *get_preferred_pack(struct pack_geometry *geometry)
+static struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry)
 {
 	uint32_t i;
 
@@ -304,9 +304,9 @@ static struct packed_git *get_preferred_pack(struct pack_geometry *geometry)
 	return NULL;
 }
 
-static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
-					    struct string_list *names,
-					    struct existing_packs *existing)
+static void pack_geometry_remove_redundant(struct pack_geometry *geometry,
+					   struct string_list *names,
+					   struct existing_packs *existing)
 {
 	const struct git_hash_algo *algop = existing->repo->hash_algo;
 	struct strbuf buf = STRBUF_INIT;
@@ -332,7 +332,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
 	strbuf_release(&buf);
 }
 
-static void free_pack_geometry(struct pack_geometry *geometry)
+static void pack_geometry_release(struct pack_geometry *geometry)
 {
 	if (!geometry)
 		return;
@@ -599,7 +599,7 @@ static int write_midx_included_packs(struct string_list *include,
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
-	struct packed_git *preferred = get_preferred_pack(geometry);
+	struct packed_git *preferred = pack_geometry_preferred_pack(geometry);
 	FILE *in;
 	int ret;
 
@@ -1063,8 +1063,8 @@ int cmd_repack(int argc,
 	if (geometry.split_factor) {
 		if (pack_everything)
 			die(_("options '%s' and '%s' cannot be used together"), "--geometric", "-A/-a");
-		init_pack_geometry(&geometry, &existing, &po_args);
-		split_pack_geometry(&geometry);
+		pack_geometry_init(&geometry, &existing, &po_args);
+		pack_geometry_split(&geometry);
 	}
 
 	prepare_pack_objects(&cmd, &po_args, packtmp);
@@ -1324,8 +1324,8 @@ int cmd_repack(int argc,
 		existing_packs_remove_redundant(&existing, packdir);
 
 		if (geometry.split_factor)
-			geometry_remove_redundant_packs(&geometry, &names,
-							&existing);
+			pack_geometry_remove_redundant(&geometry, &names,
+						       &existing);
 		if (show_progress)
 			opts |= PRUNE_PACKED_VERBOSE;
 		prune_packed_objects(opts);
@@ -1352,7 +1352,7 @@ int cmd_repack(int argc,
 	string_list_clear(&keep_pack_list, 0);
 	string_list_clear(&names, 1);
 	existing_packs_release(&existing);
-	free_pack_geometry(&geometry);
+	pack_geometry_release(&geometry);
 	for (size_t i = 0; i < midx_pack_names_nr; i++)
 		free(midx_pack_names[i]);
 	free(midx_pack_names);
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Sep 29, 2025, 23:22 UTC · re: Taylor Blau · lore

Re: [PATCH 27/49] builtin/repack.c: rename various pack_geometry functions

On Sun, Sep 28, 2025 at 06:08:54PM -0400, Taylor Blau wrote:
Show 10 quoted lines
> diff --git a/builtin/repack.c b/builtin/repack.c
> index 18c3df7200..2ce1ae3364 100644
> --- a/builtin/repack.c
> +++ b/builtin/repack.c
> @@ -115,17 +115,17 @@ struct pack_geometry {
>  	int split_factor;
>  };
>  
> -static uint32_t geometry_pack_weight(struct packed_git *p)
> +static uint32_t pack_geometry_weight(struct packed_git *p)

This rename is a bit of an oddball as we don't work on a `struct pack_geometry` here.

Show 9 quoted lines
> @@ -332,7 +332,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
>  	strbuf_release(&buf);
>  }
>  
> -static void free_pack_geometry(struct pack_geometry *geometry)
> +static void pack_geometry_release(struct pack_geometry *geometry)
>  {
>  	if (!geometry)
>  		return;

Okay, this function also gets adjusted for the new naming schema we have around free/release. Good, but it might be worth it to point it out in the commit message.

Patrick
Taylor Blau· Oct 7, 2025, 20:38 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 27/49] builtin/repack.c: rename various pack_geometry functions

On Tue, Sep 30, 2025 at 01:22:05AM +0200, Patrick Steinhardt wrote:
Show 14 quoted lines
> On Sun, Sep 28, 2025 at 06:08:54PM -0400, Taylor Blau wrote:
> > diff --git a/builtin/repack.c b/builtin/repack.c
> > index 18c3df7200..2ce1ae3364 100644
> > --- a/builtin/repack.c
> > +++ b/builtin/repack.c
> > @@ -115,17 +115,17 @@ struct pack_geometry {
> >  	int split_factor;
> >  };
> >
> > -static uint32_t geometry_pack_weight(struct packed_git *p)
> > +static uint32_t pack_geometry_weight(struct packed_git *p)
>
> This rename is a bit of an oddball as we don't work on a `struct
> pack_geometry` here.

Agreed. I made the change because the concept of a pack's "weight" in this instance is tied to the pack_geometry API here, not an intrinsic property of the pack.

Show 13 quoted lines
> > @@ -332,7 +332,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
> >  	strbuf_release(&buf);
> >  }
> >
> > -static void free_pack_geometry(struct pack_geometry *geometry)
> > +static void pack_geometry_release(struct pack_geometry *geometry)
> >  {
> >  	if (!geometry)
> >  		return;
>
> Okay, this function also gets adjusted for the new naming schema we have
> around free/release. Good, but it might be worth it to point it out in
> the commit message.

I wrote the commit message to be generic to cover "various pack_geometry functions". Did you want me to be more specific about which functions were renamed?

Thanks, Taylor

Patrick Steinhardt· Oct 8, 2025, 04:26 UTC · re: Taylor Blau · lore

Re: [PATCH 27/49] builtin/repack.c: rename various pack_geometry functions

On Tue, Oct 07, 2025 at 04:38:33PM -0400, Taylor Blau wrote:
Show 23 quoted lines
> On Tue, Sep 30, 2025 at 01:22:05AM +0200, Patrick Steinhardt wrote:
> > On Sun, Sep 28, 2025 at 06:08:54PM -0400, Taylor Blau wrote:
> > > diff --git a/builtin/repack.c b/builtin/repack.c
> > > index 18c3df7200..2ce1ae3364 100644
> > > --- a/builtin/repack.c
> > > +++ b/builtin/repack.c
> > > @@ -332,7 +332,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
> > >  	strbuf_release(&buf);
> > >  }
> > >
> > > -static void free_pack_geometry(struct pack_geometry *geometry)
> > > +static void pack_geometry_release(struct pack_geometry *geometry)
> > >  {
> > >  	if (!geometry)
> > >  		return;
> >
> > Okay, this function also gets adjusted for the new naming schema we have
> > around free/release. Good, but it might be worth it to point it out in
> > the commit message.
> 
> I wrote the commit message to be generic to cover "various pack_geometry
> functions". Did you want me to be more specific about which functions
> were renamed?

Not in general, no. But I felt that this method here was somewhat special because we not only rename to have `pack_geometry_` as prefix, but also to use the correct verb.

Patrick
Taylor Blau· Oct 10, 2025, 23:00 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 27/49] builtin/repack.c: rename various pack_geometry functions

On Wed, Oct 08, 2025 at 06:26:29AM +0200, Patrick Steinhardt wrote:
Show 28 quoted lines
> On Tue, Oct 07, 2025 at 04:38:33PM -0400, Taylor Blau wrote:
> > On Tue, Sep 30, 2025 at 01:22:05AM +0200, Patrick Steinhardt wrote:
> > > On Sun, Sep 28, 2025 at 06:08:54PM -0400, Taylor Blau wrote:
> > > > diff --git a/builtin/repack.c b/builtin/repack.c
> > > > index 18c3df7200..2ce1ae3364 100644
> > > > --- a/builtin/repack.c
> > > > +++ b/builtin/repack.c
> > > > @@ -332,7 +332,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
> > > >  	strbuf_release(&buf);
> > > >  }
> > > >
> > > > -static void free_pack_geometry(struct pack_geometry *geometry)
> > > > +static void pack_geometry_release(struct pack_geometry *geometry)
> > > >  {
> > > >  	if (!geometry)
> > > >  		return;
> > >
> > > Okay, this function also gets adjusted for the new naming schema we have
> > > around free/release. Good, but it might be worth it to point it out in
> > > the commit message.
> >
> > I wrote the commit message to be generic to cover "various pack_geometry
> > functions". Did you want me to be more specific about which functions
> > were renamed?
>
> Not in general, no. But I felt that this method here was somewhat
> special because we not only rename to have `pack_geometry_` as prefix,
> but also to use the correct verb.

Aha, sorry, I see what you're saying. I'll make that clearer in my local version, thanks for suggesting it!

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:08 UTC · re: Taylor Blau · lore

[PATCH 28/49] builtin/repack.c: pass 'pack_kept_objects' to `pack_geometry_init()`

Prepare to move pack_geometry-related APIs to their own compilation unit by passing in the static "pack_kept_objects" variable directly as a parameter to the 'pack_geometry_init()' function.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +4 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index 2ce1ae3364..60dce45556 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -136,7 +136,8 @@ static int pack_geometry_cmp(const void *va, const void *vb)
 
 static void pack_geometry_init(struct pack_geometry *geometry,
 			       struct existing_packs *existing,
-			       const struct pack_objects_args *args)
+			       const struct pack_objects_args *args,
+			       int pack_kept_objects)
 {
 	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
@@ -1063,7 +1064,8 @@ int cmd_repack(int argc,
 	if (geometry.split_factor) {
 		if (pack_everything)
 			die(_("options '%s' and '%s' cannot be used together"), "--geometric", "-A/-a");
-		pack_geometry_init(&geometry, &existing, &po_args);
+		pack_geometry_init(&geometry, &existing, &po_args,
+				   pack_kept_objects);
 		pack_geometry_split(&geometry);
 	}
 
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 29/49] builtin/repack.c: pass 'packdir' to `pack_geometry_remove_redundant()`

For similar reasons as the preceding commit, pass the "packdir" variable directly to `pack_geometry_remove_redundant()` as a parameter to the function.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +3 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index 60dce45556..9e2523d948 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -307,7 +307,8 @@ static struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geo
 
 static void pack_geometry_remove_redundant(struct pack_geometry *geometry,
 					   struct string_list *names,
-					   struct existing_packs *existing)
+					   struct existing_packs *existing,
+					   const char *packdir)
 {
 	const struct git_hash_algo *algop = existing->repo->hash_algo;
 	struct strbuf buf = STRBUF_INIT;
@@ -1327,7 +1328,7 @@ int cmd_repack(int argc,
 
 		if (geometry.split_factor)
 			pack_geometry_remove_redundant(&geometry, &names,
-						       &existing);
+						       &existing, packdir);
 		if (show_progress)
 			opts |= PRUNE_PACKED_VERBOSE;
 		prune_packed_objects(opts);
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 30/49] repack: remove pack_geometry API from the builtin

Now that the pack_geometry API is fully factored and isolated from the rest of the builtin, declare it within repack.h and move its implementation to "repack-geometry.c" as a separate component.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile          |   1 +
 builtin/repack.c  | 235 ----------------------------------------------
 meson.build       |   1 +
 repack-geometry.c | 233 +++++++++++++++++++++++++++++++++++++++++++++
 repack.h          |  20 ++++
 5 files changed, 255 insertions(+), 235 deletions(-)
 create mode 100644 repack-geometry.c
Show changes to 5 files +255 −235

Makefile, builtin/repack.c, meson.build, repack-geometry.c, repack.h

diff --git a/Makefile b/Makefile
index 0a73f8adaf..a6da12a326 100644
--- a/Makefile
+++ b/Makefile
@@ -1137,6 +1137,7 @@ LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
+LIB_OBJS += repack-geometry.o
 LIB_OBJS += repack-promisor.o
 LIB_OBJS += replace-object.o
 LIB_OBJS += repo-settings.o
diff --git a/builtin/repack.c b/builtin/repack.c
index 9e2523d948..b49e2dab9a 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,241 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-struct pack_geometry {
-	struct packed_git **pack;
-	uint32_t pack_nr, pack_alloc;
-	uint32_t split;
-
-	int split_factor;
-};
-
-static uint32_t pack_geometry_weight(struct packed_git *p)
-{
-	if (open_pack_index(p))
-		die(_("cannot open index for %s"), p->pack_name);
-	return p->num_objects;
-}
-
-static int pack_geometry_cmp(const void *va, const void *vb)
-{
-	uint32_t aw = pack_geometry_weight(*(struct packed_git **)va),
-		 bw = pack_geometry_weight(*(struct packed_git **)vb);
-
-	if (aw < bw)
-		return -1;
-	if (aw > bw)
-		return 1;
-	return 0;
-}
-
-static void pack_geometry_init(struct pack_geometry *geometry,
-			       struct existing_packs *existing,
-			       const struct pack_objects_args *args,
-			       int pack_kept_objects)
-{
-	struct packfile_store *packs = existing->repo->objects->packfiles;
-	struct packed_git *p;
-	struct strbuf buf = STRBUF_INIT;
-
-	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
-		if (args->local && !p->pack_local)
-			/*
-			 * When asked to only repack local packfiles we skip
-			 * over any packfiles that are borrowed from alternate
-			 * object directories.
-			 */
-			continue;
-
-		if (!pack_kept_objects) {
-			/*
-			 * Any pack that has its pack_keep bit set will
-			 * appear in existing->kept_packs below, but
-			 * this saves us from doing a more expensive
-			 * check.
-			 */
-			if (p->pack_keep)
-				continue;
-
-			/*
-			 * The pack may be kept via the --keep-pack
-			 * option; check 'existing->kept_packs' to
-			 * determine whether to ignore it.
-			 */
-			strbuf_reset(&buf);
-			strbuf_addstr(&buf, pack_basename(p));
-			strbuf_strip_suffix(&buf, ".pack");
-
-			if (string_list_has_string(&existing->kept_packs, buf.buf))
-				continue;
-		}
-		if (p->is_cruft)
-			continue;
-
-		ALLOC_GROW(geometry->pack,
-			   geometry->pack_nr + 1,
-			   geometry->pack_alloc);
-
-		geometry->pack[geometry->pack_nr] = p;
-		geometry->pack_nr++;
-	}
-
-	QSORT(geometry->pack, geometry->pack_nr, pack_geometry_cmp);
-	strbuf_release(&buf);
-}
-
-static void pack_geometry_split(struct pack_geometry *geometry)
-{
-	uint32_t i;
-	uint32_t split;
-	off_t total_size = 0;
-
-	if (!geometry->pack_nr) {
-		geometry->split = geometry->pack_nr;
-		return;
-	}
-
-	/*
-	 * First, count the number of packs (in descending order of size) which
-	 * already form a geometric progression.
-	 */
-	for (i = geometry->pack_nr - 1; i > 0; i--) {
-		struct packed_git *ours = geometry->pack[i];
-		struct packed_git *prev = geometry->pack[i - 1];
-
-		if (unsigned_mult_overflows(geometry->split_factor,
-					    pack_geometry_weight(prev)))
-			die(_("pack %s too large to consider in geometric "
-			      "progression"),
-			    prev->pack_name);
-
-		if (pack_geometry_weight(ours) <
-		    geometry->split_factor * pack_geometry_weight(prev))
-			break;
-	}
-
-	split = i;
-
-	if (split) {
-		/*
-		 * Move the split one to the right, since the top element in the
-		 * last-compared pair can't be in the progression. Only do this
-		 * when we split in the middle of the array (otherwise if we got
-		 * to the end, then the split is in the right place).
-		 */
-		split++;
-	}
-
-	/*
-	 * Then, anything to the left of 'split' must be in a new pack. But,
-	 * creating that new pack may cause packs in the heavy half to no longer
-	 * form a geometric progression.
-	 *
-	 * Compute an expected size of the new pack, and then determine how many
-	 * packs in the heavy half need to be joined into it (if any) to restore
-	 * the geometric progression.
-	 */
-	for (i = 0; i < split; i++) {
-		struct packed_git *p = geometry->pack[i];
-
-		if (unsigned_add_overflows(total_size, pack_geometry_weight(p)))
-			die(_("pack %s too large to roll up"), p->pack_name);
-		total_size += pack_geometry_weight(p);
-	}
-	for (i = split; i < geometry->pack_nr; i++) {
-		struct packed_git *ours = geometry->pack[i];
-
-		if (unsigned_mult_overflows(geometry->split_factor,
-					    total_size))
-			die(_("pack %s too large to roll up"), ours->pack_name);
-
-		if (pack_geometry_weight(ours) <
-		    geometry->split_factor * total_size) {
-			if (unsigned_add_overflows(total_size,
-						   pack_geometry_weight(ours)))
-				die(_("pack %s too large to roll up"),
-				    ours->pack_name);
-
-			split++;
-			total_size += pack_geometry_weight(ours);
-		} else
-			break;
-	}
-
-	geometry->split = split;
-}
-
-static struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry)
-{
-	uint32_t i;
-
-	if (!geometry) {
-		/*
-		 * No geometry means either an all-into-one repack (in which
-		 * case there is only one pack left and it is the largest) or an
-		 * incremental one.
-		 *
-		 * If repacking incrementally, then we could check the size of
-		 * all packs to determine which should be preferred, but leave
-		 * this for later.
-		 */
-		return NULL;
-	}
-	if (geometry->split == geometry->pack_nr)
-		return NULL;
-
-	/*
-	 * The preferred pack is the largest pack above the split line. In
-	 * other words, it is the largest pack that does not get rolled up in
-	 * the geometric repack.
-	 */
-	for (i = geometry->pack_nr; i > geometry->split; i--)
-		/*
-		 * A pack that is not local would never be included in a
-		 * multi-pack index. We thus skip over any non-local packs.
-		 */
-		if (geometry->pack[i - 1]->pack_local)
-			return geometry->pack[i - 1];
-
-	return NULL;
-}
-
-static void pack_geometry_remove_redundant(struct pack_geometry *geometry,
-					   struct string_list *names,
-					   struct existing_packs *existing,
-					   const char *packdir)
-{
-	const struct git_hash_algo *algop = existing->repo->hash_algo;
-	struct strbuf buf = STRBUF_INIT;
-	uint32_t i;
-
-	for (i = 0; i < geometry->split; i++) {
-		struct packed_git *p = geometry->pack[i];
-		if (string_list_has_string(names, hash_to_hex_algop(p->hash,
-								    algop)))
-			continue;
-
-		strbuf_reset(&buf);
-		strbuf_addstr(&buf, pack_basename(p));
-		strbuf_strip_suffix(&buf, ".pack");
-
-		if ((p->pack_keep) ||
-		    (string_list_has_string(&existing->kept_packs, buf.buf)))
-			continue;
-
-		repack_remove_redundant_pack(existing->repo, packdir, buf.buf);
-	}
-
-	strbuf_release(&buf);
-}
-
-static void pack_geometry_release(struct pack_geometry *geometry)
-{
-	if (!geometry)
-		return;
-
-	free(geometry->pack);
-}
-
 static int midx_has_unknown_packs(char **midx_pack_names,
 				  size_t midx_pack_names_nr,
 				  struct string_list *include,
diff --git a/meson.build b/meson.build
index ca5603b456..c929ad6f7e 100644
--- a/meson.build
+++ b/meson.build
@@ -463,6 +463,7 @@ libgit_sources = [
   'reftable/writer.c',
   'remote.c',
   'repack.c',
+  'repack-geometry.c',
   'repack-promisor.c',
   'replace-object.c',
   'repo-settings.c',
diff --git a/repack-geometry.c b/repack-geometry.c
new file mode 100644
index 0000000000..a879f2fe49
--- /dev/null
+++ b/repack-geometry.c
@@ -0,0 +1,233 @@
+#define DISABLE_SIGN_COMPARE_WARNINGS
+
+#include "git-compat-util.h"
+#include "repack.h"
+#include "hex.h"
+#include "packfile.h"
+
+static uint32_t pack_geometry_weight(struct packed_git *p)
+{
+	if (open_pack_index(p))
+		die(_("cannot open index for %s"), p->pack_name);
+	return p->num_objects;
+}
+
+static int pack_geometry_cmp(const void *va, const void *vb)
+{
+	uint32_t aw = pack_geometry_weight(*(struct packed_git **)va),
+		 bw = pack_geometry_weight(*(struct packed_git **)vb);
+
+	if (aw < bw)
+		return -1;
+	if (aw > bw)
+		return 1;
+	return 0;
+}
+
+void pack_geometry_init(struct pack_geometry *geometry,
+			struct existing_packs *existing,
+			const struct pack_objects_args *args,
+			int pack_kept_objects)
+{
+	struct packfile_store *packs = existing->repo->objects->packfiles;
+	struct packed_git *p;
+	struct strbuf buf = STRBUF_INIT;
+
+	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
+		if (args->local && !p->pack_local)
+			/*
+			 * When asked to only repack local packfiles we skip
+			 * over any packfiles that are borrowed from alternate
+			 * object directories.
+			 */
+			continue;
+
+		if (!pack_kept_objects) {
+			/*
+			 * Any pack that has its pack_keep bit set will
+			 * appear in existing->kept_packs below, but
+			 * this saves us from doing a more expensive
+			 * check.
+			 */
+			if (p->pack_keep)
+				continue;
+
+			/*
+			 * The pack may be kept via the --keep-pack
+			 * option; check 'existing->kept_packs' to
+			 * determine whether to ignore it.
+			 */
+			strbuf_reset(&buf);
+			strbuf_addstr(&buf, pack_basename(p));
+			strbuf_strip_suffix(&buf, ".pack");
+
+			if (string_list_has_string(&existing->kept_packs, buf.buf))
+				continue;
+		}
+		if (p->is_cruft)
+			continue;
+
+		ALLOC_GROW(geometry->pack,
+			   geometry->pack_nr + 1,
+			   geometry->pack_alloc);
+
+		geometry->pack[geometry->pack_nr] = p;
+		geometry->pack_nr++;
+	}
+
+	QSORT(geometry->pack, geometry->pack_nr, pack_geometry_cmp);
+	strbuf_release(&buf);
+}
+
+void pack_geometry_split(struct pack_geometry *geometry)
+{
+	uint32_t i;
+	uint32_t split;
+	off_t total_size = 0;
+
+	if (!geometry->pack_nr) {
+		geometry->split = geometry->pack_nr;
+		return;
+	}
+
+	/*
+	 * First, count the number of packs (in descending order of size) which
+	 * already form a geometric progression.
+	 */
+	for (i = geometry->pack_nr - 1; i > 0; i--) {
+		struct packed_git *ours = geometry->pack[i];
+		struct packed_git *prev = geometry->pack[i - 1];
+
+		if (unsigned_mult_overflows(geometry->split_factor,
+					    pack_geometry_weight(prev)))
+			die(_("pack %s too large to consider in geometric "
+			      "progression"),
+			    prev->pack_name);
+
+		if (pack_geometry_weight(ours) <
+		    geometry->split_factor * pack_geometry_weight(prev))
+			break;
+	}
+
+	split = i;
+
+	if (split) {
+		/*
+		 * Move the split one to the right, since the top element in the
+		 * last-compared pair can't be in the progression. Only do this
+		 * when we split in the middle of the array (otherwise if we got
+		 * to the end, then the split is in the right place).
+		 */
+		split++;
+	}
+
+	/*
+	 * Then, anything to the left of 'split' must be in a new pack. But,
+	 * creating that new pack may cause packs in the heavy half to no longer
+	 * form a geometric progression.
+	 *
+	 * Compute an expected size of the new pack, and then determine how many
+	 * packs in the heavy half need to be joined into it (if any) to restore
+	 * the geometric progression.
+	 */
+	for (i = 0; i < split; i++) {
+		struct packed_git *p = geometry->pack[i];
+
+		if (unsigned_add_overflows(total_size, pack_geometry_weight(p)))
+			die(_("pack %s too large to roll up"), p->pack_name);
+		total_size += pack_geometry_weight(p);
+	}
+	for (i = split; i < geometry->pack_nr; i++) {
+		struct packed_git *ours = geometry->pack[i];
+
+		if (unsigned_mult_overflows(geometry->split_factor,
+					    total_size))
+			die(_("pack %s too large to roll up"), ours->pack_name);
+
+		if (pack_geometry_weight(ours) <
+		    geometry->split_factor * total_size) {
+			if (unsigned_add_overflows(total_size,
+						   pack_geometry_weight(ours)))
+				die(_("pack %s too large to roll up"),
+				    ours->pack_name);
+
+			split++;
+			total_size += pack_geometry_weight(ours);
+		} else
+			break;
+	}
+
+	geometry->split = split;
+}
+
+struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry)
+{
+	uint32_t i;
+
+	if (!geometry) {
+		/*
+		 * No geometry means either an all-into-one repack (in which
+		 * case there is only one pack left and it is the largest) or an
+		 * incremental one.
+		 *
+		 * If repacking incrementally, then we could check the size of
+		 * all packs to determine which should be preferred, but leave
+		 * this for later.
+		 */
+		return NULL;
+	}
+	if (geometry->split == geometry->pack_nr)
+		return NULL;
+
+	/*
+	 * The preferred pack is the largest pack above the split line. In
+	 * other words, it is the largest pack that does not get rolled up in
+	 * the geometric repack.
+	 */
+	for (i = geometry->pack_nr; i > geometry->split; i--)
+		/*
+		 * A pack that is not local would never be included in a
+		 * multi-pack index. We thus skip over any non-local packs.
+		 */
+		if (geometry->pack[i - 1]->pack_local)
+			return geometry->pack[i - 1];
+
+	return NULL;
+}
+
+void pack_geometry_remove_redundant(struct pack_geometry *geometry,
+				    struct string_list *names,
+				    struct existing_packs *existing,
+				    const char *packdir)
+{
+	const struct git_hash_algo *algop = existing->repo->hash_algo;
+	struct strbuf buf = STRBUF_INIT;
+	uint32_t i;
+
+	for (i = 0; i < geometry->split; i++) {
+		struct packed_git *p = geometry->pack[i];
+		if (string_list_has_string(names, hash_to_hex_algop(p->hash,
+								    algop)))
+			continue;
+
+		strbuf_reset(&buf);
+		strbuf_addstr(&buf, pack_basename(p));
+		strbuf_strip_suffix(&buf, ".pack");
+
+		if ((p->pack_keep) ||
+		    (string_list_has_string(&existing->kept_packs, buf.buf)))
+			continue;
+
+		repack_remove_redundant_pack(existing->repo, packdir, buf.buf);
+	}
+
+	strbuf_release(&buf);
+}
+
+void pack_geometry_release(struct pack_geometry *geometry)
+{
+	if (!geometry)
+		return;
+
+	free(geometry->pack);
+}
diff --git a/repack.h b/repack.h
index 19dc4fd738..cea7969ae4 100644
--- a/repack.h
+++ b/repack.h
@@ -78,4 +78,24 @@ void repack_promisor_objects(struct repository *repo,
 			     const struct pack_objects_args *args,
 			     struct string_list *names, const char *packtmp);
 
+struct pack_geometry {
+	struct packed_git **pack;
+	uint32_t pack_nr, pack_alloc;
+	uint32_t split;
+
+	int split_factor;
+};
+
+void pack_geometry_init(struct pack_geometry *geometry,
+			struct existing_packs *existing,
+			const struct pack_objects_args *args,
+			int pack_kept_objects);
+void pack_geometry_split(struct pack_geometry *geometry);
+struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry);
+void pack_geometry_remove_redundant(struct pack_geometry *geometry,
+				    struct string_list *names,
+				    struct existing_packs *existing,
+				    const char *packdir);
+void pack_geometry_release(struct pack_geometry *geometry);
+
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 31/49] builtin/repack.c: remove ref snapshotting from builtin

When writing a MIDX, 'git repack' takes a snapshot of the repository's references and writes the result out to a file, which it then passes to 'git multi-pack-index write' via the '--refs-snapshot'.

This is done in order to make bitmap selections with respect to what we are packing, thus avoiding a race where an incoming reference update causes us to try and write a bitmap for a commit not present in the MIDX.

Extract this functionality out into a new repack-midx.c compilation unit, and expose the necessary functions via the repack.h API.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile         |  1 +
 builtin/repack.c | 68 ------------------------------------------
 meson.build      |  1 +
 repack-midx.c    | 77 ++++++++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  4 +++
 5 files changed, 83 insertions(+), 68 deletions(-)
 create mode 100644 repack-midx.c
Show changes to 5 files +83 −68

Makefile, builtin/repack.c, meson.build, repack-midx.c, repack.h

diff --git a/Makefile b/Makefile
index a6da12a326..3360743afb 100644
--- a/Makefile
+++ b/Makefile
@@ -1138,6 +1138,7 @@ LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
 LIB_OBJS += repack-geometry.o
+LIB_OBJS += repack-midx.o
 LIB_OBJS += repack-promisor.o
 LIB_OBJS += replace-object.o
 LIB_OBJS += repo-settings.o
diff --git a/builtin/repack.c b/builtin/repack.c
index b49e2dab9a..5a1f1cb562 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -178,74 +178,6 @@ static int midx_has_unknown_packs(char **midx_pack_names,
 	return 0;
 }
 
-struct midx_snapshot_ref_data {
-	struct repository *repo;
-	struct tempfile *f;
-	struct oidset seen;
-	int preferred;
-};
-
-static int midx_snapshot_ref_one(const char *refname UNUSED,
-				 const char *referent UNUSED,
-				 const struct object_id *oid,
-				 int flag UNUSED, void *_data)
-{
-	struct midx_snapshot_ref_data *data = _data;
-	struct object_id peeled;
-
-	if (!peel_iterated_oid(data->repo, oid, &peeled))
-		oid = &peeled;
-
-	if (oidset_insert(&data->seen, oid))
-		return 0; /* already seen */
-
-	if (odb_read_object_info(data->repo->objects, oid, NULL) != OBJ_COMMIT)
-		return 0;
-
-	fprintf(data->f->fp, "%s%s\n", data->preferred ? "+" : "",
-		oid_to_hex(oid));
-
-	return 0;
-}
-
-static void midx_snapshot_refs(struct repository *repo, struct tempfile *f)
-{
-	struct midx_snapshot_ref_data data;
-	const struct string_list *preferred = bitmap_preferred_tips(repo);
-
-	data.repo = repo;
-	data.f = f;
-	data.preferred = 0;
-	oidset_init(&data.seen, 0);
-
-	if (!fdopen_tempfile(f, "w"))
-		 die(_("could not open tempfile %s for writing"),
-		     get_tempfile_path(f));
-
-	if (preferred) {
-		struct string_list_item *item;
-
-		data.preferred = 1;
-		for_each_string_list_item(item, preferred)
-			refs_for_each_ref_in(get_main_ref_store(repo),
-					     item->string,
-					     midx_snapshot_ref_one, &data);
-		data.preferred = 0;
-	}
-
-	refs_for_each_ref(get_main_ref_store(repo),
-			  midx_snapshot_ref_one, &data);
-
-	if (close_tempfile_gently(f)) {
-		int save_errno = errno;
-		delete_tempfile(&f);
-		errno = save_errno;
-		die_errno(_("could not close refs snapshot tempfile"));
-	}
-
-	oidset_clear(&data.seen);
-}
-
 static void midx_included_packs(struct string_list *include,
 				struct existing_packs *existing,
 				char **midx_pack_names,
diff --git a/meson.build b/meson.build
index c929ad6f7e..42171d1851 100644
--- a/meson.build
+++ b/meson.build
@@ -464,6 +464,7 @@ libgit_sources = [
   'remote.c',
   'repack.c',
   'repack-geometry.c',
+  'repack-midx.c',
   'repack-promisor.c',
   'replace-object.c',
   'repo-settings.c',
diff --git a/repack-midx.c b/repack-midx.c
new file mode 100644
index 0000000000..354df729a5
--- /dev/null
+++ b/repack-midx.c
@@ -0,0 +1,77 @@
+#include "git-compat-util.h"
+#include "repack.h"
+#include "hash.h"
+#include "hex.h"
+#include "odb.h"
+#include "oidset.h"
+#include "pack-bitmap.h"
+#include "refs.h"
+#include "tempfile.h"
+
+struct midx_snapshot_ref_data {
+	struct repository *repo;
+	struct tempfile *f;
+	struct oidset seen;
+	int preferred;
+};
+
+static int midx_snapshot_ref_one(const char *refname UNUSED,
+				 const char *referent UNUSED,
+				 const struct object_id *oid,
+				 int flag UNUSED, void *_data)
+{
+	struct midx_snapshot_ref_data *data = _data;
+	struct object_id peeled;
+
+	if (!peel_iterated_oid(data->repo, oid, &peeled))
+		oid = &peeled;
+
+	if (oidset_insert(&data->seen, oid))
+		return 0; /* already seen */
+
+	if (odb_read_object_info(data->repo->objects, oid, NULL) != OBJ_COMMIT)
+		return 0;
+
+	fprintf(data->f->fp, "%s%s\n", data->preferred ? "+" : "",
+		oid_to_hex(oid));
+
+	return 0;
+}
+
+void midx_snapshot_refs(struct repository *repo, struct tempfile *f)
+{
+	struct midx_snapshot_ref_data data;
+	const struct string_list *preferred = bitmap_preferred_tips(repo);
+
+	data.repo = repo;
+	data.f = f;
+	data.preferred = 0;
+	oidset_init(&data.seen, 0);
+
+	if (!fdopen_tempfile(f, "w"))
+		 die(_("could not open tempfile %s for writing"),
+		     get_tempfile_path(f));
+
+	if (preferred) {
+		struct string_list_item *item;
+
+		data.preferred = 1;
+		for_each_string_list_item(item, preferred)
+			refs_for_each_ref_in(get_main_ref_store(repo),
+					     item->string,
+					     midx_snapshot_ref_one, &data);
+		data.preferred = 0;
+	}
+
+	refs_for_each_ref(get_main_ref_store(repo),
+			  midx_snapshot_ref_one, &data);
+
+	if (close_tempfile_gently(f)) {
+		int save_errno = errno;
+		delete_tempfile(&f);
+		errno = save_errno;
+		die_errno(_("could not close refs snapshot tempfile"));
+	}
+
+	oidset_clear(&data.seen);
+}
diff --git a/repack.h b/repack.h
index cea7969ae4..803e129224 100644
--- a/repack.h
+++ b/repack.h
@@ -98,4 +98,8 @@ void pack_geometry_remove_redundant(struct pack_geometry *geometry,
 				    const char *packdir);
 void pack_geometry_release(struct pack_geometry *geometry);
 
+struct tempfile;
+
+void midx_snapshot_refs(struct repository *repo, struct tempfile *f);
+
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 32/49] builtin/repack.c: extract opts struct for 'write_midx_included_packs()'

The function 'write_midx_included_packs()', which is responsible for writing a new MIDX with a given set of included packs, currently takes a list of six arguments.

In order to extract this function out of the builtin, we have to pass in a few additional parameters, like 'midx_must_contain_cruft' and 'packdir', which are currently declared as static variables within the builtin/repack.c compilation unit.

Instead of adding additional parameters to `write_midx_included_packs()` extract out an "opts" struct that names these parameters, and pass a pointer to that, making it less cumbersome to add additional parameters.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 52 +++++++++++++++++++++++++++++++-----------------
 1 file changed, 34 insertions(+), 18 deletions(-)
Show changes to builtin/repack.c +34 −18
diff --git a/builtin/repack.c b/builtin/repack.c
index 5a1f1cb562..e4b8aa2c6f 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,6 +107,17 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
+struct repack_write_midx_opts {
+	struct string_list *include;
+	struct pack_geometry *geometry;
+	struct string_list *names;
+	const char *refs_snapshot;
+	const char *packdir;
+	int show_progress;
+	int write_bitmaps;
+	int midx_must_contain_cruft;
+};
+
 static int midx_has_unknown_packs(char **midx_pack_names,
 				  size_t midx_pack_names_nr,
 				  struct string_list *include,
@@ -290,19 +301,15 @@ static void midx_included_packs(struct string_list *include,
 	strbuf_release(&buf);
 }
 
-static int write_midx_included_packs(struct string_list *include,
-				     struct pack_geometry *geometry,
-				     struct string_list *names,
-				     const char *refs_snapshot,
-				     int show_progress, int write_bitmaps)
+static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
-	struct packed_git *preferred = pack_geometry_preferred_pack(geometry);
+	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
 	FILE *in;
 	int ret;
 
-	if (!include->nr)
+	if (!opts->include->nr)
 		return 0;
 
 	cmd.in = -1;
@@ -311,18 +318,18 @@ static int write_midx_included_packs(struct string_list *include,
 	strvec_push(&cmd.args, "multi-pack-index");
 	strvec_pushl(&cmd.args, "write", "--stdin-packs", NULL);
 
-	if (show_progress)
+	if (opts->show_progress)
 		strvec_push(&cmd.args, "--progress");
 	else
 		strvec_push(&cmd.args, "--no-progress");
 
-	if (write_bitmaps)
+	if (opts->write_bitmaps)
 		strvec_push(&cmd.args, "--bitmap");
 
 	if (preferred)
 		strvec_pushf(&cmd.args, "--preferred-pack=%s",
 			     pack_basename(preferred));
-	else if (names->nr) {
+	else if (opts->names->nr) {
 		/* The largest pack was repacked, meaning that either
 		 * one or two packs exist depending on whether the
 		 * repository has a cruft pack or not.
@@ -335,7 +342,7 @@ static int write_midx_included_packs(struct string_list *include,
 		 * `--max-pack-size` was given, but any one of them
 		 * will suffice, so pick the first one.)
 		 */
-		for_each_string_list_item(item, names) {
+		for_each_string_list_item(item, opts->names) {
 			struct generated_pack *pack = item->util;
 			if (generated_pack_has_ext(pack, ".mtimes"))
 				continue;
@@ -355,15 +362,16 @@ static int write_midx_included_packs(struct string_list *include,
 		;
 	}
 
-	if (refs_snapshot)
-		strvec_pushf(&cmd.args, "--refs-snapshot=%s", refs_snapshot);
+	if (opts->refs_snapshot)
+		strvec_pushf(&cmd.args, "--refs-snapshot=%s",
+			     opts->refs_snapshot);
 
 	ret = start_command(&cmd);
 	if (ret)
 		return ret;
 
 	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, include)
+	for_each_string_list_item(item, opts->include)
 		fprintf(in, "%s\n", item->string);
 	fclose(in);
 
@@ -1001,15 +1009,23 @@ int cmd_repack(int argc,
 
 	if (write_midx) {
 		struct string_list include = STRING_LIST_INIT_DUP;
+		struct repack_write_midx_opts opts = {
+			.include = &include,
+			.geometry = &geometry,
+			.names = &names,
+			.refs_snapshot = refs_snapshot ? get_tempfile_path(refs_snapshot) : NULL,
+			.packdir = packdir,
+			.show_progress = show_progress,
+			.write_bitmaps = write_bitmaps > 0,
+			.midx_must_contain_cruft = midx_must_contain_cruft
+		};
 		midx_included_packs(&include, &existing, midx_pack_names,
 				    midx_pack_names_nr, &names, &geometry);
 
-		ret = write_midx_included_packs(&include, &geometry, &names,
-						refs_snapshot ? get_tempfile_path(refs_snapshot) : NULL,
-						show_progress, write_bitmaps > 0);
+		ret = write_midx_included_packs(&opts);
 
 		if (!ret && write_bitmaps)
-			remove_redundant_bitmaps(&include, packdir);
+			remove_redundant_bitmaps(&include, opts.packdir);
 
 		string_list_clear(&include, 0);
 
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 33/49] builtin/repack.c: use a string_list for 'midx_pack_names'

When writing a new MIDX, repack must determine whether or not there are any packs in the MIDX it is replacing (if one exists) that are not somehow represented in the new MIDX (e.g., either by preserving the pack verbatim, or rolling it up as part of a geometric repack, etc.).

In order to do this, it keeps track of a list of pack names from the MIDX present in the repository at the start of the repack operation. Since we manipulate and close the object store, we cannot rely on the repository's in-core representation of the MIDX, since this is subject to change and/or go away.

When this behavior was introduced in 5ee86c273b (repack: exclude cruft pack(s) from the MIDX where possible, 2025-06-23), we maintained an array of character pointers instead of using a convenience API, such as string-list.h.

Store the list of MIDX pack names in a string_list, thereby reducing the number of parameters we have to pass to `midx_has_unknown_packs()`.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 40 +++++++++++++++++-----------------------
 1 file changed, 17 insertions(+), 23 deletions(-)
Show changes to builtin/repack.c +17 −23
diff --git a/builtin/repack.c b/builtin/repack.c
index e4b8aa2c6f..8ae56b05e9 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -118,18 +118,17 @@ struct repack_write_midx_opts {
 	int midx_must_contain_cruft;
 };
 
-static int midx_has_unknown_packs(char **midx_pack_names,
-				  size_t midx_pack_names_nr,
+static int midx_has_unknown_packs(struct string_list *midx_pack_names,
 				  struct string_list *include,
 				  struct pack_geometry *geometry,
 				  struct existing_packs *existing)
 {
-	size_t i;
+	struct string_list_item *item;
 
 	string_list_sort(include);
 
-	for (i = 0; i < midx_pack_names_nr; i++) {
-		const char *pack_name = midx_pack_names[i];
+	for_each_string_list_item(item, midx_pack_names) {
+		const char *pack_name = item->string;
 
 		/*
 		 * Determine whether or not each MIDX'd pack from the existing
@@ -191,8 +190,7 @@ static int midx_has_unknown_packs(char **midx_pack_names,
 
 static void midx_included_packs(struct string_list *include,
 				struct existing_packs *existing,
-				char **midx_pack_names,
-				size_t midx_pack_names_nr,
+				struct string_list *midx_pack_names,
 				struct string_list *names,
 				struct pack_geometry *geometry)
 {
@@ -247,8 +245,8 @@ static void midx_included_packs(struct string_list *include,
 	}
 
 	if (midx_must_contain_cruft ||
-	    midx_has_unknown_packs(midx_pack_names, midx_pack_names_nr,
-				   include, geometry, existing)) {
+	    midx_has_unknown_packs(midx_pack_names, include, geometry,
+				   existing)) {
 		/*
 		 * If there are one or more unknown pack(s) present (see
 		 * midx_has_unknown_packs() for what makes a pack
@@ -606,13 +604,12 @@ int cmd_repack(int argc,
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
 	struct string_list names = STRING_LIST_INIT_DUP;
+	struct string_list midx_pack_names = STRING_LIST_INIT_DUP;
 	struct existing_packs existing = EXISTING_PACKS_INIT;
 	struct pack_geometry geometry = { 0 };
 	struct tempfile *refs_snapshot = NULL;
 	int i, ret;
 	int show_progress;
-	char **midx_pack_names = NULL;
-	size_t midx_pack_names_nr = 0;
 
 	/* variables to be filled by option parsing */
 	struct repack_config_ctx config_ctx;
@@ -985,13 +982,12 @@ int cmd_repack(int argc,
 		struct multi_pack_index *m =
 			get_multi_pack_index(repo->objects->sources);
 
-		ALLOC_ARRAY(midx_pack_names,
-			    m->num_packs + m->num_packs_in_base);
-
-		for (; m; m = m->base_midx)
-			for (uint32_t i = 0; i < m->num_packs; i++)
-				midx_pack_names[midx_pack_names_nr++] =
-					xstrdup(m->pack_names[i]);
+		for (; m; m = m->base_midx) {
+			for (uint32_t i = 0; i < m->num_packs; i++) {
+				string_list_append(&midx_pack_names,
+						   m->pack_names[i]);
+			}
+		}
 	}
 
 	close_object_store(repo->objects);
@@ -1019,8 +1015,8 @@ int cmd_repack(int argc,
 			.write_bitmaps = write_bitmaps > 0,
 			.midx_must_contain_cruft = midx_must_contain_cruft
 		};
-		midx_included_packs(&include, &existing, midx_pack_names,
-				    midx_pack_names_nr, &names, &geometry);
+		midx_included_packs(&include, &existing, &midx_pack_names,
+				    &names, &geometry);
 
 		ret = write_midx_included_packs(&opts);
 
@@ -1067,11 +1063,9 @@ int cmd_repack(int argc,
 cleanup:
 	string_list_clear(&keep_pack_list, 0);
 	string_list_clear(&names, 1);
+	string_list_clear(&midx_pack_names, 0);
 	existing_packs_release(&existing);
 	pack_geometry_release(&geometry);
-	for (size_t i = 0; i < midx_pack_names_nr; i++)
-		free(midx_pack_names[i]);
-	free(midx_pack_names);
 	pack_objects_args_release(&po_args);
 	pack_objects_args_release(&cruft_po_args);
 
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 34/49] repack: keep track of MIDX pack names using existing_packs

Instead of storing the list of MIDX pack names separately, let's inline it into the existing_packs struct, further reducing the number of parameters we have to pass around.

This amounts to adding a new string_list to the existing_packs struct, and populating it via `existing_packs_collect()`. This is fairly straightforward to do, since we are already looping over all packs, all we need to do is:

    if (p->multi_pack_index)
        string_list_append(&existing->midx_packs, pack_basename(p));

Note, however, that this check *must* come before other conditions where we discard and do not keep track of a pack, including the condition "if (!p->pack_local)" immediately below. This is because the existing routine which collects MIDX pack names does so blindly, and does not discard, for example, non-local packs.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 26 ++++----------------------
 repack.c         |  5 +++++
 repack.h         |  1 +
 3 files changed, 10 insertions(+), 22 deletions(-)
Show changes to 3 files +10 −22

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 8ae56b05e9..251dd08b0a 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -118,8 +118,7 @@ struct repack_write_midx_opts {
 	int midx_must_contain_cruft;
 };
 
-static int midx_has_unknown_packs(struct string_list *midx_pack_names,
-				  struct string_list *include,
+static int midx_has_unknown_packs(struct string_list *include,
 				  struct pack_geometry *geometry,
 				  struct existing_packs *existing)
 {
@@ -127,7 +126,7 @@ static int midx_has_unknown_packs(struct string_list *midx_pack_names,
 
 	string_list_sort(include);
 
-	for_each_string_list_item(item, midx_pack_names) {
+	for_each_string_list_item(item, &existing->midx_packs) {
 		const char *pack_name = item->string;
 
 		/*
@@ -190,7 +189,6 @@ static int midx_has_unknown_packs(struct string_list *midx_pack_names,
 
 static void midx_included_packs(struct string_list *include,
 				struct existing_packs *existing,
-				struct string_list *midx_pack_names,
 				struct string_list *names,
 				struct pack_geometry *geometry)
 {
@@ -245,8 +243,7 @@ static void midx_included_packs(struct string_list *include,
 	}
 
 	if (midx_must_contain_cruft ||
-	    midx_has_unknown_packs(midx_pack_names, include, geometry,
-				   existing)) {
+	    midx_has_unknown_packs(include, geometry, existing)) {
 		/*
 		 * If there are one or more unknown pack(s) present (see
 		 * midx_has_unknown_packs() for what makes a pack
@@ -604,7 +601,6 @@ int cmd_repack(int argc,
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
 	struct string_list names = STRING_LIST_INIT_DUP;
-	struct string_list midx_pack_names = STRING_LIST_INIT_DUP;
 	struct existing_packs existing = EXISTING_PACKS_INIT;
 	struct pack_geometry geometry = { 0 };
 	struct tempfile *refs_snapshot = NULL;
@@ -978,18 +974,6 @@ int cmd_repack(int argc,
 
 	string_list_sort(&names);
 
-	if (get_multi_pack_index(repo->objects->sources)) {
-		struct multi_pack_index *m =
-			get_multi_pack_index(repo->objects->sources);
-
-		for (; m; m = m->base_midx) {
-			for (uint32_t i = 0; i < m->num_packs; i++) {
-				string_list_append(&midx_pack_names,
-						   m->pack_names[i]);
-			}
-		}
-	}
-
 	close_object_store(repo->objects);
 
 	/*
@@ -1015,8 +999,7 @@ int cmd_repack(int argc,
 			.write_bitmaps = write_bitmaps > 0,
 			.midx_must_contain_cruft = midx_must_contain_cruft
 		};
-		midx_included_packs(&include, &existing, &midx_pack_names,
-				    &names, &geometry);
+		midx_included_packs(&include, &existing, &names, &geometry);
 
 		ret = write_midx_included_packs(&opts);
 
@@ -1063,7 +1046,6 @@ int cmd_repack(int argc,
 cleanup:
 	string_list_clear(&keep_pack_list, 0);
 	string_list_clear(&names, 1);
-	string_list_clear(&midx_pack_names, 0);
 	existing_packs_release(&existing);
 	pack_geometry_release(&geometry);
 	pack_objects_args_release(&po_args);
diff --git a/repack.c b/repack.c
index d8afdd352d..1d485e0112 100644
--- a/repack.c
+++ b/repack.c
@@ -80,6 +80,9 @@ void existing_packs_collect(struct existing_packs *existing,
 		size_t i;
 		const char *base;
 
+		if (p->multi_pack_index)
+			string_list_append(&existing->midx_packs,
+					    pack_basename(p));
 		if (!p->pack_local)
 			continue;
 
@@ -104,6 +107,7 @@ void existing_packs_collect(struct existing_packs *existing,
 	string_list_sort(&existing->kept_packs);
 	string_list_sort(&existing->non_kept_packs);
 	string_list_sort(&existing->cruft_packs);
+	string_list_sort(&existing->midx_packs);
 	strbuf_release(&buf);
 }
 
@@ -220,6 +224,7 @@ void existing_packs_release(struct existing_packs *existing)
 	string_list_clear(&existing->kept_packs, 0);
 	string_list_clear(&existing->non_kept_packs, 0);
 	string_list_clear(&existing->cruft_packs, 0);
+	string_list_clear(&existing->midx_packs, 0);
 }
 
 static struct {
diff --git a/repack.h b/repack.h
index 803e129224..6aa5b4e0f0 100644
--- a/repack.h
+++ b/repack.h
@@ -40,6 +40,7 @@ struct existing_packs {
 	struct string_list kept_packs;
 	struct string_list non_kept_packs;
 	struct string_list cruft_packs;
+	struct string_list midx_packs;
 };
 
 #define EXISTING_PACKS_INIT { \
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 35/49] builtin/repack.c: reorder `remove_redundant_bitmaps()`

The next commit will inline the call to `remove_redundant_bitmaps()` into `write_midx_included_packs()`. Reorder these two functions to avoid a forward declaration to `remove_redundant_bitmaps()`.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 58 ++++++++++++++++++++++++------------------------
 1 file changed, 29 insertions(+), 29 deletions(-)
Show changes to builtin/repack.c +29 −29
diff --git a/builtin/repack.c b/builtin/repack.c
index 251dd08b0a..957a9b5f9e 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -296,6 +296,35 @@ static void midx_included_packs(struct string_list *include,
 	strbuf_release(&buf);
 }
 
+static void remove_redundant_bitmaps(struct string_list *include,
+				     const char *packdir)
+{
+	struct strbuf path = STRBUF_INIT;
+	struct string_list_item *item;
+	size_t packdir_len;
+
+	strbuf_addstr(&path, packdir);
+	strbuf_addch(&path, '/');
+	packdir_len = path.len;
+
+	/*
+	 * Remove any pack bitmaps corresponding to packs which are now
+	 * included in the MIDX.
+	 */
+	for_each_string_list_item(item, include) {
+		strbuf_addstr(&path, item->string);
+		strbuf_strip_suffix(&path, ".idx");
+		strbuf_addstr(&path, ".bitmap");
+
+		if (unlink(path.buf) && errno != ENOENT)
+			warning_errno(_("could not remove stale bitmap: %s"),
+				      path.buf);
+
+		strbuf_setlen(&path, packdir_len);
+	}
+	strbuf_release(&path);
+}
+
 static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
@@ -373,35 +402,6 @@ static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 	return finish_command(&cmd);
 }
 
-static void remove_redundant_bitmaps(struct string_list *include,
-				     const char *packdir)
-{
-	struct strbuf path = STRBUF_INIT;
-	struct string_list_item *item;
-	size_t packdir_len;
-
-	strbuf_addstr(&path, packdir);
-	strbuf_addch(&path, '/');
-	packdir_len = path.len;
-
-	/*
-	 * Remove any pack bitmaps corresponding to packs which are now
-	 * included in the MIDX.
-	 */
-	for_each_string_list_item(item, include) {
-		strbuf_addstr(&path, item->string);
-		strbuf_strip_suffix(&path, ".idx");
-		strbuf_addstr(&path, ".bitmap");
-
-		if (unlink(path.buf) && errno != ENOENT)
-			warning_errno(_("could not remove stale bitmap: %s"),
-				      path.buf);
-
-		strbuf_setlen(&path, packdir_len);
-	}
-	strbuf_release(&path);
-}
-
 static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 				   struct child_process *cmd,
 				   struct string_list *names,
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 36/49] builtin/repack.c: inline `remove_redundant_bitmaps()`

After writing a new MIDX, the repack command removes any bitmaps belonging to packs which were written into the MIDX.

This is currently done in a separate function outside of `write_midx_included_packs()`, which forces the caller to keep track of the set of packs written into the MIDX.

Prepare to no longer require the caller to keep track of such information by inlining the clean-up into `write_midx_included_packs()`. Future commits will make the caller oblivious to the set of packs included in the MIDX altogether.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)
Show changes to builtin/repack.c +8 −7
diff --git a/builtin/repack.c b/builtin/repack.c
index 957a9b5f9e..b55c8934e8 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -331,10 +331,10 @@ static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 	struct string_list_item *item;
 	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
 	FILE *in;
-	int ret;
+	int ret = 0;
 
 	if (!opts->include->nr)
-		return 0;
+		goto done;
 
 	cmd.in = -1;
 	cmd.git_cmd = 1;
@@ -392,14 +392,18 @@ static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 
 	ret = start_command(&cmd);
 	if (ret)
-		return ret;
+		goto done;
 
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, opts->include)
 		fprintf(in, "%s\n", item->string);
 	fclose(in);
 
-	return finish_command(&cmd);
+	ret = finish_command(&cmd);
+done:
+	if (!ret && opts->write_bitmaps)
+		remove_redundant_bitmaps(opts->include, opts->packdir);
+	return ret;
 }
 
 static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
@@ -1003,9 +1007,6 @@ int cmd_repack(int argc,
 
 		ret = write_midx_included_packs(&opts);
 
-		if (!ret && write_bitmaps)
-			remove_redundant_bitmaps(&include, opts.packdir);
-
 		string_list_clear(&include, 0);
 
 		if (ret)
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 37/49] builtin/repack.c: pass `repack_write_midx_opts` to `midx_included_packs`

Instead of passing individual parameters (in this case, "existing", "names", and "geometry") to `midx_included_packs()`, pass a pointer to a `repack_write_midx_opts` structure instead.

Besides reducing the number of parameters necessary to call the `midx_included_packs` function, this refactoring sets us up nicely to inline the call to `midx_included_packs()` into `write_midx_included_packs()`, thus making the caller (in this case, `cmd_repack()`) oblivious to the set of packs being written into the MIDX.

In order to do this, `repack_write_midx_opts` has to keep track of the set of existing packs, so add an additional field to point to that set.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)
Show changes to builtin/repack.c +8 −5
diff --git a/builtin/repack.c b/builtin/repack.c
index b55c8934e8..bd3034a4f0 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -108,6 +108,7 @@ static int repack_config(const char *var, const char *value,
 }
 
 struct repack_write_midx_opts {
+	struct existing_packs *existing;
 	struct string_list *include;
 	struct pack_geometry *geometry;
 	struct string_list *names;
@@ -188,10 +189,11 @@ static int midx_has_unknown_packs(struct string_list *include,
 }
 
 static void midx_included_packs(struct string_list *include,
-				struct existing_packs *existing,
-				struct string_list *names,
-				struct pack_geometry *geometry)
+				struct repack_write_midx_opts *opts)
 {
+	struct existing_packs *existing = opts->existing;
+	struct pack_geometry *geometry = opts->geometry;
+	struct string_list *names = opts->names;
 	struct string_list_item *item;
 	struct strbuf buf = STRBUF_INIT;
 
@@ -242,7 +244,7 @@ static void midx_included_packs(struct string_list *include,
 		}
 	}
 
-	if (midx_must_contain_cruft ||
+	if (opts->midx_must_contain_cruft ||
 	    midx_has_unknown_packs(include, geometry, existing)) {
 		/*
 		 * If there are one or more unknown pack(s) present (see
@@ -994,6 +996,7 @@ int cmd_repack(int argc,
 	if (write_midx) {
 		struct string_list include = STRING_LIST_INIT_DUP;
 		struct repack_write_midx_opts opts = {
+			.existing = &existing,
 			.include = &include,
 			.geometry = &geometry,
 			.names = &names,
@@ -1003,7 +1006,7 @@ int cmd_repack(int argc,
 			.write_bitmaps = write_bitmaps > 0,
 			.midx_must_contain_cruft = midx_must_contain_cruft
 		};
-		midx_included_packs(&include, &existing, &names, &geometry);
+		midx_included_packs(&include, &opts);
 
 		ret = write_midx_included_packs(&opts);
 
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 38/49] builtin/repack.c: inline packs within `write_midx_included_packs()`

To write a MIDX at the end of a repack operation, 'git repack' presently computes the set of packs to write into the MIDX, before invoking `write_midx_included_packs()` with a `string_list` containing those packs.

The logic for computing which packs are supposed to appear in the resulting MIDX is within `midx_included_packs()`, where it is aware of details like which cruft pack(s) were written/combined, if/how we did a geometric repack, etc.

Computing this list ourselves before providing it to the sole function to make use of that list `write_midx_included_packs()` is somewhat awkward. In the future, repack will learn how to write incremental MIDXs, which will use a very different pack selection routine.

Instead of doing something like:
    struct string_list included_packs = STRING_LIST_INIT_DUP;
    if (incremental) {
        midx_incremental_included_packs(&included_packs, ...):
        write_midx_incremental_included_packs(&included_packs, ...);
    } else {
        midx_included_packs(&included_packs, ...):
        write_midx_included_packs(&included_packs, ...);
    }

in the future, let's have each function which writes a MIDX be responsible for itself computing the list of included packs. Inline the declaration and initialization of `included_packs` into the `write_midx_included_packs()` function itself, and repeat that pattern in the future when we introduce new ways to write MIDXs.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 17 ++++++++---------
 1 file changed, 8 insertions(+), 9 deletions(-)
Show changes to builtin/repack.c +8 −9
diff --git a/builtin/repack.c b/builtin/repack.c
index bd3034a4f0..e2acba6312 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -109,7 +109,6 @@ static int repack_config(const char *var, const char *value,
 
 struct repack_write_midx_opts {
 	struct existing_packs *existing;
-	struct string_list *include;
 	struct pack_geometry *geometry;
 	struct string_list *names;
 	const char *refs_snapshot;
@@ -330,12 +329,14 @@ static void remove_redundant_bitmaps(struct string_list *include,
 static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
+	struct string_list include = STRING_LIST_INIT_DUP;
 	struct string_list_item *item;
 	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
 	FILE *in;
 	int ret = 0;
 
-	if (!opts->include->nr)
+	midx_included_packs(&include, opts);
+	if (!include.nr)
 		goto done;
 
 	cmd.in = -1;
@@ -397,14 +398,17 @@ static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 		goto done;
 
 	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, opts->include)
+	for_each_string_list_item(item, &include)
 		fprintf(in, "%s\n", item->string);
 	fclose(in);
 
 	ret = finish_command(&cmd);
 done:
 	if (!ret && opts->write_bitmaps)
-		remove_redundant_bitmaps(opts->include, opts->packdir);
+		remove_redundant_bitmaps(&include, opts->packdir);
+
+	string_list_clear(&include, 0);
+
 	return ret;
 }
 
@@ -994,10 +998,8 @@ int cmd_repack(int argc,
 		existing_packs_mark_for_deletion(&existing, &names);
 
 	if (write_midx) {
-		struct string_list include = STRING_LIST_INIT_DUP;
 		struct repack_write_midx_opts opts = {
 			.existing = &existing,
-			.include = &include,
 			.geometry = &geometry,
 			.names = &names,
 			.refs_snapshot = refs_snapshot ? get_tempfile_path(refs_snapshot) : NULL,
@@ -1006,12 +1008,9 @@ int cmd_repack(int argc,
 			.write_bitmaps = write_bitmaps > 0,
 			.midx_must_contain_cruft = midx_must_contain_cruft
 		};
-		midx_included_packs(&include, &opts);
 
 		ret = write_midx_included_packs(&opts);
 
-		string_list_clear(&include, 0);
-
 		if (ret)
 			goto cleanup;
 	}
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Oct 10, 2025, 12:28 UTC · re: Taylor Blau · lore

Re: [PATCH 38/49] builtin/repack.c: inline packs within `write_midx_included_packs()`

On Sun, Sep 28, 2025 at 06:09:40PM -0400, Taylor Blau wrote:
Show 27 quoted lines
> To write a MIDX at the end of a repack operation, 'git repack' presently
> computes the set of packs to write into the MIDX, before invoking
> `write_midx_included_packs()` with a `string_list` containing those
> packs.
> 
> The logic for computing which packs are supposed to appear in the
> resulting MIDX is within `midx_included_packs()`, where it is aware of
> details like which cruft pack(s) were written/combined, if/how we did a
> geometric repack, etc.
> 
> Computing this list ourselves before providing it to the sole function
> to make use of that list `write_midx_included_packs()` is somewhat
> awkward. In the future, repack will learn how to write incremental
> MIDXs, which will use a very different pack selection routine.
> 
> Instead of doing something like:
> 
>     struct string_list included_packs = STRING_LIST_INIT_DUP;
>     if (incremental) {
>         midx_incremental_included_packs(&included_packs, ...):
>         write_midx_incremental_included_packs(&included_packs, ...);
>     } else {
>         midx_included_packs(&included_packs, ...):
>         write_midx_included_packs(&included_packs, ...);
>     }
> 
> in the future, let's have each function which writes a MIDX be
s/writes/that &/
Other than that this patch LGTM.
Patrick
Taylor Blau· Oct 10, 2025, 23:02 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 38/49] builtin/repack.c: inline packs within `write_midx_included_packs()`

On Fri, Oct 10, 2025 at 02:28:09PM +0200, Patrick Steinhardt wrote:
Show 30 quoted lines
> On Sun, Sep 28, 2025 at 06:09:40PM -0400, Taylor Blau wrote:
> > To write a MIDX at the end of a repack operation, 'git repack' presently
> > computes the set of packs to write into the MIDX, before invoking
> > `write_midx_included_packs()` with a `string_list` containing those
> > packs.
> >
> > The logic for computing which packs are supposed to appear in the
> > resulting MIDX is within `midx_included_packs()`, where it is aware of
> > details like which cruft pack(s) were written/combined, if/how we did a
> > geometric repack, etc.
> >
> > Computing this list ourselves before providing it to the sole function
> > to make use of that list `write_midx_included_packs()` is somewhat
> > awkward. In the future, repack will learn how to write incremental
> > MIDXs, which will use a very different pack selection routine.
> >
> > Instead of doing something like:
> >
> >     struct string_list included_packs = STRING_LIST_INIT_DUP;
> >     if (incremental) {
> >         midx_incremental_included_packs(&included_packs, ...):
> >         write_midx_incremental_included_packs(&included_packs, ...);
> >     } else {
> >         midx_included_packs(&included_packs, ...):
> >         write_midx_included_packs(&included_packs, ...);
> >     }
> >
> > in the future, let's have each function which writes a MIDX be
>
> s/writes/that &/

Good catch, "that writes" is definitely grammatically correct, but I think the substitution is s/which/that/, not s/writes/that &/.

Thanks, Taylor

Jeff King· Oct 15, 2025, 10:26 UTC · re: Taylor Blau · lore

Re: [PATCH 38/49] builtin/repack.c: inline packs within `write_midx_included_packs()`

On Sun, Sep 28, 2025 at 06:09:40PM -0400, Taylor Blau wrote:
Show 26 quoted lines
> The logic for computing which packs are supposed to appear in the
> resulting MIDX is within `midx_included_packs()`, where it is aware of
> details like which cruft pack(s) were written/combined, if/how we did a
> geometric repack, etc.
> 
> Computing this list ourselves before providing it to the sole function
> to make use of that list `write_midx_included_packs()` is somewhat
> awkward. In the future, repack will learn how to write incremental
> MIDXs, which will use a very different pack selection routine.
> 
> Instead of doing something like:
> 
>     struct string_list included_packs = STRING_LIST_INIT_DUP;
>     if (incremental) {
>         midx_incremental_included_packs(&included_packs, ...):
>         write_midx_incremental_included_packs(&included_packs, ...);
>     } else {
>         midx_included_packs(&included_packs, ...):
>         write_midx_included_packs(&included_packs, ...);
>     }
> 
> in the future, let's have each function which writes a MIDX be
> responsible for itself computing the list of included packs. Inline the
> declaration and initialization of `included_packs` into the
> `write_midx_included_packs()` function itself, and repeat that pattern
> in the future when we introduce new ways to write MIDXs.

OK. I thought at first this was doing something complicated, but it really is just: move the call to midx_included_packs() down into the function which uses it, rather than doing it in the caller.

Without another caller we can't judge this, but I'll take your word that it helps eventually.

The patch itself looks like it should be a noop with respect to behavior.

-Peff
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 39/49] repack: 'write_midx_included_packs' API from the builtin

Now that we have sufficiently cleaned up the write_midx_included_packs() function, we can move it (along with the struct repack_write_midx_opts) out of the builtin, and into the repack.h header.

Since this function (and the static ones that it depends on) are MIDX-specific details of the repacking process, move them to the repack-midx.c compilation unit instead of the general repack.c one.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 305 -----------------------------------------------
 repack-midx.c    | 295 +++++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  12 ++
 3 files changed, 307 insertions(+), 305 deletions(-)
Show changes to 3 files +307 −305

builtin/repack.c, repack-midx.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index e2acba6312..5fed79e826 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,311 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-struct repack_write_midx_opts {
-	struct existing_packs *existing;
-	struct pack_geometry *geometry;
-	struct string_list *names;
-	const char *refs_snapshot;
-	const char *packdir;
-	int show_progress;
-	int write_bitmaps;
-	int midx_must_contain_cruft;
-};
-
-static int midx_has_unknown_packs(struct string_list *include,
-				  struct pack_geometry *geometry,
-				  struct existing_packs *existing)
-{
-	struct string_list_item *item;
-
-	string_list_sort(include);
-
-	for_each_string_list_item(item, &existing->midx_packs) {
-		const char *pack_name = item->string;
-
-		/*
-		 * Determine whether or not each MIDX'd pack from the existing
-		 * MIDX (if any) is represented in the new MIDX. For each pack
-		 * in the MIDX, it must either be:
-		 *
-		 *  - In the "include" list of packs to be included in the new
-		 *    MIDX. Note this function is called before the include
-		 *    list is populated with any cruft pack(s).
-		 *
-		 *  - Below the geometric split line (if using pack geometry),
-		 *    indicating that the pack won't be included in the new
-		 *    MIDX, but its contents were rolled up as part of the
-		 *    geometric repack.
-		 *
-		 *  - In the existing non-kept packs list (if not using pack
-		 *    geometry), and marked as non-deleted.
-		 */
-		if (string_list_has_string(include, pack_name)) {
-			continue;
-		} else if (geometry) {
-			struct strbuf buf = STRBUF_INIT;
-			uint32_t j;
-
-			for (j = 0; j < geometry->split; j++) {
-				strbuf_reset(&buf);
-				strbuf_addstr(&buf, pack_basename(geometry->pack[j]));
-				strbuf_strip_suffix(&buf, ".pack");
-				strbuf_addstr(&buf, ".idx");
-
-				if (!strcmp(pack_name, buf.buf)) {
-					strbuf_release(&buf);
-					break;
-				}
-			}
-
-			strbuf_release(&buf);
-
-			if (j < geometry->split)
-				continue;
-		} else {
-			struct string_list_item *item;
-
-			item = string_list_lookup(&existing->non_kept_packs,
-						  pack_name);
-			if (item && !existing_pack_is_marked_for_deletion(item))
-				continue;
-		}
-
-		/*
-		 * If we got to this point, the MIDX includes some pack that we
-		 * don't know about.
-		 */
-		return 1;
-	}
-
-	return 0;
-}
-
-static void midx_included_packs(struct string_list *include,
-				struct repack_write_midx_opts *opts)
-{
-	struct existing_packs *existing = opts->existing;
-	struct pack_geometry *geometry = opts->geometry;
-	struct string_list *names = opts->names;
-	struct string_list_item *item;
-	struct strbuf buf = STRBUF_INIT;
-
-	for_each_string_list_item(item, &existing->kept_packs) {
-		strbuf_reset(&buf);
-		strbuf_addf(&buf, "%s.idx", item->string);
-		string_list_insert(include, buf.buf);
-	}
-
-	for_each_string_list_item(item, names) {
-		strbuf_reset(&buf);
-		strbuf_addf(&buf, "pack-%s.idx", item->string);
-		string_list_insert(include, buf.buf);
-	}
-
-	if (geometry->split_factor) {
-		uint32_t i;
-
-		for (i = geometry->split; i < geometry->pack_nr; i++) {
-			struct packed_git *p = geometry->pack[i];
-
-			/*
-			 * The multi-pack index never refers to packfiles part
-			 * of an alternate object database, so we skip these.
-			 * While git-multi-pack-index(1) would silently ignore
-			 * them anyway, this allows us to skip executing the
-			 * command completely when we have only non-local
-			 * packfiles.
-			 */
-			if (!p->pack_local)
-				continue;
-
-			strbuf_reset(&buf);
-			strbuf_addstr(&buf, pack_basename(p));
-			strbuf_strip_suffix(&buf, ".pack");
-			strbuf_addstr(&buf, ".idx");
-
-			string_list_insert(include, buf.buf);
-		}
-	} else {
-		for_each_string_list_item(item, &existing->non_kept_packs) {
-			if (existing_pack_is_marked_for_deletion(item))
-				continue;
-
-			strbuf_reset(&buf);
-			strbuf_addf(&buf, "%s.idx", item->string);
-			string_list_insert(include, buf.buf);
-		}
-	}
-
-	if (opts->midx_must_contain_cruft ||
-	    midx_has_unknown_packs(include, geometry, existing)) {
-		/*
-		 * If there are one or more unknown pack(s) present (see
-		 * midx_has_unknown_packs() for what makes a pack
-		 * "unknown") in the MIDX before the repack, keep them
-		 * as they may be required to form a reachability
-		 * closure if the MIDX is bitmapped.
-		 *
-		 * For example, a cruft pack can be required to form a
-		 * reachability closure if the MIDX is bitmapped and one
-		 * or more of the bitmap's selected commits reaches a
-		 * once-cruft object that was later made reachable.
-		 */
-		for_each_string_list_item(item, &existing->cruft_packs) {
-			/*
-			 * When doing a --geometric repack, there is no
-			 * need to check for deleted packs, since we're
-			 * by definition not doing an ALL_INTO_ONE
-			 * repack (hence no packs will be deleted).
-			 * Otherwise we must check for and exclude any
-			 * packs which are enqueued for deletion.
-			 *
-			 * So we could omit the conditional below in the
-			 * --geometric case, but doing so is unnecessary
-			 *  since no packs are marked as pending
-			 *  deletion (since we only call
-			 *  `existing_packs_mark_for_deletion()` when
-			 *  doing an all-into-one repack).
-			 */
-			if (existing_pack_is_marked_for_deletion(item))
-				continue;
-
-			strbuf_reset(&buf);
-			strbuf_addf(&buf, "%s.idx", item->string);
-			string_list_insert(include, buf.buf);
-		}
-	} else {
-		/*
-		 * Modern versions of Git (with the appropriate
-		 * configuration setting) will write new copies of
-		 * once-cruft objects when doing a --geometric repack.
-		 *
-		 * If the MIDX has no cruft pack, new packs written
-		 * during a --geometric repack will not rely on the
-		 * cruft pack to form a reachability closure, so we can
-		 * avoid including them in the MIDX in that case.
-		 */
-		;
-	}
-
-	strbuf_release(&buf);
-}
-
-static void remove_redundant_bitmaps(struct string_list *include,
-				     const char *packdir)
-{
-	struct strbuf path = STRBUF_INIT;
-	struct string_list_item *item;
-	size_t packdir_len;
-
-	strbuf_addstr(&path, packdir);
-	strbuf_addch(&path, '/');
-	packdir_len = path.len;
-
-	/*
-	 * Remove any pack bitmaps corresponding to packs which are now
-	 * included in the MIDX.
-	 */
-	for_each_string_list_item(item, include) {
-		strbuf_addstr(&path, item->string);
-		strbuf_strip_suffix(&path, ".idx");
-		strbuf_addstr(&path, ".bitmap");
-
-		if (unlink(path.buf) && errno != ENOENT)
-			warning_errno(_("could not remove stale bitmap: %s"),
-				      path.buf);
-
-		strbuf_setlen(&path, packdir_len);
-	}
-	strbuf_release(&path);
-}
-
-static int write_midx_included_packs(struct repack_write_midx_opts *opts)
-{
-	struct child_process cmd = CHILD_PROCESS_INIT;
-	struct string_list include = STRING_LIST_INIT_DUP;
-	struct string_list_item *item;
-	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
-	FILE *in;
-	int ret = 0;
-
-	midx_included_packs(&include, opts);
-	if (!include.nr)
-		goto done;
-
-	cmd.in = -1;
-	cmd.git_cmd = 1;
-
-	strvec_push(&cmd.args, "multi-pack-index");
-	strvec_pushl(&cmd.args, "write", "--stdin-packs", NULL);
-
-	if (opts->show_progress)
-		strvec_push(&cmd.args, "--progress");
-	else
-		strvec_push(&cmd.args, "--no-progress");
-
-	if (opts->write_bitmaps)
-		strvec_push(&cmd.args, "--bitmap");
-
-	if (preferred)
-		strvec_pushf(&cmd.args, "--preferred-pack=%s",
-			     pack_basename(preferred));
-	else if (opts->names->nr) {
-		/* The largest pack was repacked, meaning that either
-		 * one or two packs exist depending on whether the
-		 * repository has a cruft pack or not.
-		 *
-		 * Select the non-cruft one as preferred to encourage
-		 * pack-reuse among packs containing reachable objects
-		 * over unreachable ones.
-		 *
-		 * (Note we could write multiple packs here if
-		 * `--max-pack-size` was given, but any one of them
-		 * will suffice, so pick the first one.)
-		 */
-		for_each_string_list_item(item, opts->names) {
-			struct generated_pack *pack = item->util;
-			if (generated_pack_has_ext(pack, ".mtimes"))
-				continue;
-
-			strvec_pushf(&cmd.args, "--preferred-pack=pack-%s.pack",
-				     item->string);
-			break;
-		}
-	} else {
-		/*
-		 * No packs were kept, and no packs were written. The
-		 * only thing remaining are .keep packs (unless
-		 * --pack-kept-objects was given).
-		 *
-		 * Set the `--preferred-pack` arbitrarily here.
-		 */
-		;
-	}
-
-	if (opts->refs_snapshot)
-		strvec_pushf(&cmd.args, "--refs-snapshot=%s",
-			     opts->refs_snapshot);
-
-	ret = start_command(&cmd);
-	if (ret)
-		goto done;
-
-	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, &include)
-		fprintf(in, "%s\n", item->string);
-	fclose(in);
-
-	ret = finish_command(&cmd);
-done:
-	if (!ret && opts->write_bitmaps)
-		remove_redundant_bitmaps(&include, opts->packdir);
-
-	string_list_clear(&include, 0);
-
-	return ret;
-}
-
 static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 				   struct child_process *cmd,
 				   struct string_list *names,
diff --git a/repack-midx.c b/repack-midx.c
index 354df729a5..6f6202c5bc 100644
--- a/repack-midx.c
+++ b/repack-midx.c
@@ -6,6 +6,7 @@
 #include "oidset.h"
 #include "pack-bitmap.h"
 #include "refs.h"
+#include "run-command.h"
 #include "tempfile.h"
 
 struct midx_snapshot_ref_data {
@@ -75,3 +76,297 @@ void midx_snapshot_refs(struct repository *repo, struct tempfile *f)
 
 	oidset_clear(&data.seen);
 }
+
+static int midx_has_unknown_packs(struct string_list *include,
+				  struct pack_geometry *geometry,
+				  struct existing_packs *existing)
+{
+	struct string_list_item *item;
+
+	string_list_sort(include);
+
+	for_each_string_list_item(item, &existing->midx_packs) {
+		const char *pack_name = item->string;
+
+		/*
+		 * Determine whether or not each MIDX'd pack from the existing
+		 * MIDX (if any) is represented in the new MIDX. For each pack
+		 * in the MIDX, it must either be:
+		 *
+		 *  - In the "include" list of packs to be included in the new
+		 *    MIDX. Note this function is called before the include
+		 *    list is populated with any cruft pack(s).
+		 *
+		 *  - Below the geometric split line (if using pack geometry),
+		 *    indicating that the pack won't be included in the new
+		 *    MIDX, but its contents were rolled up as part of the
+		 *    geometric repack.
+		 *
+		 *  - In the existing non-kept packs list (if not using pack
+		 *    geometry), and marked as non-deleted.
+		 */
+		if (string_list_has_string(include, pack_name)) {
+			continue;
+		} else if (geometry) {
+			struct strbuf buf = STRBUF_INIT;
+			uint32_t j;
+
+			for (j = 0; j < geometry->split; j++) {
+				strbuf_reset(&buf);
+				strbuf_addstr(&buf, pack_basename(geometry->pack[j]));
+				strbuf_strip_suffix(&buf, ".pack");
+				strbuf_addstr(&buf, ".idx");
+
+				if (!strcmp(pack_name, buf.buf)) {
+					strbuf_release(&buf);
+					break;
+				}
+			}
+
+			strbuf_release(&buf);
+
+			if (j < geometry->split)
+				continue;
+		} else {
+			struct string_list_item *item;
+
+			item = string_list_lookup(&existing->non_kept_packs,
+						  pack_name);
+			if (item && !existing_pack_is_marked_for_deletion(item))
+				continue;
+		}
+
+		/*
+		 * If we got to this point, the MIDX includes some pack that we
+		 * don't know about.
+		 */
+		return 1;
+	}
+
+	return 0;
+}
+
+static void midx_included_packs(struct string_list *include,
+				struct repack_write_midx_opts *opts)
+{
+	struct existing_packs *existing = opts->existing;
+	struct pack_geometry *geometry = opts->geometry;
+	struct string_list *names = opts->names;
+	struct string_list_item *item;
+	struct strbuf buf = STRBUF_INIT;
+
+	for_each_string_list_item(item, &existing->kept_packs) {
+		strbuf_reset(&buf);
+		strbuf_addf(&buf, "%s.idx", item->string);
+		string_list_insert(include, buf.buf);
+	}
+
+	for_each_string_list_item(item, names) {
+		strbuf_reset(&buf);
+		strbuf_addf(&buf, "pack-%s.idx", item->string);
+		string_list_insert(include, buf.buf);
+	}
+
+	if (geometry->split_factor) {
+		uint32_t i;
+
+		for (i = geometry->split; i < geometry->pack_nr; i++) {
+			struct packed_git *p = geometry->pack[i];
+
+			/*
+			 * The multi-pack index never refers to packfiles part
+			 * of an alternate object database, so we skip these.
+			 * While git-multi-pack-index(1) would silently ignore
+			 * them anyway, this allows us to skip executing the
+			 * command completely when we have only non-local
+			 * packfiles.
+			 */
+			if (!p->pack_local)
+				continue;
+
+			strbuf_reset(&buf);
+			strbuf_addstr(&buf, pack_basename(p));
+			strbuf_strip_suffix(&buf, ".pack");
+			strbuf_addstr(&buf, ".idx");
+
+			string_list_insert(include, buf.buf);
+		}
+	} else {
+		for_each_string_list_item(item, &existing->non_kept_packs) {
+			if (existing_pack_is_marked_for_deletion(item))
+				continue;
+
+			strbuf_reset(&buf);
+			strbuf_addf(&buf, "%s.idx", item->string);
+			string_list_insert(include, buf.buf);
+		}
+	}
+
+	if (opts->midx_must_contain_cruft ||
+	    midx_has_unknown_packs(include, geometry, existing)) {
+		/*
+		 * If there are one or more unknown pack(s) present (see
+		 * midx_has_unknown_packs() for what makes a pack
+		 * "unknown") in the MIDX before the repack, keep them
+		 * as they may be required to form a reachability
+		 * closure if the MIDX is bitmapped.
+		 *
+		 * For example, a cruft pack can be required to form a
+		 * reachability closure if the MIDX is bitmapped and one
+		 * or more of the bitmap's selected commits reaches a
+		 * once-cruft object that was later made reachable.
+		 */
+		for_each_string_list_item(item, &existing->cruft_packs) {
+			/*
+			 * When doing a --geometric repack, there is no
+			 * need to check for deleted packs, since we're
+			 * by definition not doing an ALL_INTO_ONE
+			 * repack (hence no packs will be deleted).
+			 * Otherwise we must check for and exclude any
+			 * packs which are enqueued for deletion.
+			 *
+			 * So we could omit the conditional below in the
+			 * --geometric case, but doing so is unnecessary
+			 *  since no packs are marked as pending
+			 *  deletion (since we only call
+			 *  `existing_packs_mark_for_deletion()` when
+			 *  doing an all-into-one repack).
+			 */
+			if (existing_pack_is_marked_for_deletion(item))
+				continue;
+
+			strbuf_reset(&buf);
+			strbuf_addf(&buf, "%s.idx", item->string);
+			string_list_insert(include, buf.buf);
+		}
+	} else {
+		/*
+		 * Modern versions of Git (with the appropriate
+		 * configuration setting) will write new copies of
+		 * once-cruft objects when doing a --geometric repack.
+		 *
+		 * If the MIDX has no cruft pack, new packs written
+		 * during a --geometric repack will not rely on the
+		 * cruft pack to form a reachability closure, so we can
+		 * avoid including them in the MIDX in that case.
+		 */
+		;
+	}
+
+	strbuf_release(&buf);
+}
+
+static void remove_redundant_bitmaps(struct string_list *include,
+				     const char *packdir)
+{
+	struct strbuf path = STRBUF_INIT;
+	struct string_list_item *item;
+	size_t packdir_len;
+
+	strbuf_addstr(&path, packdir);
+	strbuf_addch(&path, '/');
+	packdir_len = path.len;
+
+	/*
+	 * Remove any pack bitmaps corresponding to packs which are now
+	 * included in the MIDX.
+	 */
+	for_each_string_list_item(item, include) {
+		strbuf_addstr(&path, item->string);
+		strbuf_strip_suffix(&path, ".idx");
+		strbuf_addstr(&path, ".bitmap");
+
+		if (unlink(path.buf) && errno != ENOENT)
+			warning_errno(_("could not remove stale bitmap: %s"),
+				      path.buf);
+
+		strbuf_setlen(&path, packdir_len);
+	}
+	strbuf_release(&path);
+}
+
+int write_midx_included_packs(struct repack_write_midx_opts *opts)
+{
+	struct child_process cmd = CHILD_PROCESS_INIT;
+	struct string_list include = STRING_LIST_INIT_DUP;
+	struct string_list_item *item;
+	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
+	FILE *in;
+	int ret = 0;
+
+	midx_included_packs(&include, opts);
+	if (!include.nr)
+		goto done;
+
+	cmd.in = -1;
+	cmd.git_cmd = 1;
+
+	strvec_push(&cmd.args, "multi-pack-index");
+	strvec_pushl(&cmd.args, "write", "--stdin-packs", NULL);
+
+	if (opts->show_progress)
+		strvec_push(&cmd.args, "--progress");
+	else
+		strvec_push(&cmd.args, "--no-progress");
+
+	if (opts->write_bitmaps)
+		strvec_push(&cmd.args, "--bitmap");
+
+	if (preferred)
+		strvec_pushf(&cmd.args, "--preferred-pack=%s",
+			     pack_basename(preferred));
+	else if (opts->names->nr) {
+		/* The largest pack was repacked, meaning that either
+		 * one or two packs exist depending on whether the
+		 * repository has a cruft pack or not.
+		 *
+		 * Select the non-cruft one as preferred to encourage
+		 * pack-reuse among packs containing reachable objects
+		 * over unreachable ones.
+		 *
+		 * (Note we could write multiple packs here if
+		 * `--max-pack-size` was given, but any one of them
+		 * will suffice, so pick the first one.)
+		 */
+		for_each_string_list_item(item, opts->names) {
+			struct generated_pack *pack = item->util;
+			if (generated_pack_has_ext(pack, ".mtimes"))
+				continue;
+
+			strvec_pushf(&cmd.args, "--preferred-pack=pack-%s.pack",
+				     item->string);
+			break;
+		}
+	} else {
+		/*
+		 * No packs were kept, and no packs were written. The
+		 * only thing remaining are .keep packs (unless
+		 * --pack-kept-objects was given).
+		 *
+		 * Set the `--preferred-pack` arbitrarily here.
+		 */
+		;
+	}
+
+	if (opts->refs_snapshot)
+		strvec_pushf(&cmd.args, "--refs-snapshot=%s",
+			     opts->refs_snapshot);
+
+	ret = start_command(&cmd);
+	if (ret)
+		goto done;
+
+	in = xfdopen(cmd.in, "w");
+	for_each_string_list_item(item, &include)
+		fprintf(in, "%s\n", item->string);
+	fclose(in);
+
+	ret = finish_command(&cmd);
+done:
+	if (!ret && opts->write_bitmaps)
+		remove_redundant_bitmaps(&include, opts->packdir);
+
+	string_list_clear(&include, 0);
+
+	return ret;
+}
diff --git a/repack.h b/repack.h
index 6aa5b4e0f0..25a31ac0a0 100644
--- a/repack.h
+++ b/repack.h
@@ -101,6 +101,18 @@ void pack_geometry_release(struct pack_geometry *geometry);
 
 struct tempfile;
 
+struct repack_write_midx_opts {
+	struct existing_packs *existing;
+	struct pack_geometry *geometry;
+	struct string_list *names;
+	const char *refs_snapshot;
+	const char *packdir;
+	int show_progress;
+	int write_bitmaps;
+	int midx_must_contain_cruft;
+};
+
 void midx_snapshot_refs(struct repository *repo, struct tempfile *f);
+int write_midx_included_packs(struct repack_write_midx_opts *opts);
 
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 40/49] builtin/repack.c: introduce `struct write_pack_opts`

There are various functions within the 'repack' builtin which are responsible for writing different kinds of packs. They include:

 - `static int write_filtered_pack(...)`
 - `static int write_cruft_pack(...)`

as well as the function `finish_pack_objects_cmd()`, which is responsible for finalizing a new pack write, and recording the checksum of its contents in the 'names' list.

Both of these `write_` functions have a few things in common. They both take a pointer to the 'pack_objects_args' struct, as well as a pair of character pointers for `destination` and `pack_prefix`.

Instead of repeating those arguments for each function, let's extract an options struct called "write_pack_opts" which has these three parameters as member fields. While we're at it, add fields for "packdir," and "packtmp", both of which are static variables within the builtin, and need to be read from within these two functions.

This will shorten the list of parameters that callers have to provide to `write_filtered_pack()`, avoid ambiguity when passing multiple variables of the same type, and provide a unified interface for the two functions mentioned earlier.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 28 +++++++++++++++-------------
 repack.h         |  8 ++++++++
 2 files changed, 23 insertions(+), 13 deletions(-)
Show changes to 2 files +23 −13

builtin/repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 5fed79e826..6df7c88085 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -138,9 +138,7 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 	return finish_command(cmd);
 }
 
-static int write_filtered_pack(const struct pack_objects_args *args,
-			       const char *destination,
-			       const char *pack_prefix,
+static int write_filtered_pack(struct write_pack_opts *opts,
 			       struct existing_packs *existing,
 			       struct string_list *names)
 {
@@ -150,9 +148,9 @@ static int write_filtered_pack(const struct pack_objects_args *args,
 	int ret;
 	const char *caret;
 	const char *scratch;
-	int local = skip_prefix(destination, packdir, &scratch);
+	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
 
-	prepare_pack_objects(&cmd, args, destination);
+	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
 
 	strvec_push(&cmd.args, "--stdin-packs");
 
@@ -175,7 +173,7 @@ static int write_filtered_pack(const struct pack_objects_args *args,
 	 */
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, names)
-		fprintf(in, "^%s-%s.pack\n", pack_prefix, item->string);
+		fprintf(in, "^%s-%s.pack\n", opts->pack_prefix, item->string);
 	for_each_string_list_item(item, &existing->non_kept_packs)
 		fprintf(in, "%s.pack\n", item->string);
 	for_each_string_list_item(item, &existing->cruft_packs)
@@ -665,14 +663,18 @@ int cmd_repack(int argc,
 	}
 
 	if (po_args.filter_options.choice) {
-		if (!filter_to)
-			filter_to = packtmp;
+		struct write_pack_opts opts = {
+			.po_args = &po_args,
+			.destination = filter_to,
+			.pack_prefix = find_pack_prefix(packdir, packtmp),
+			.packdir = packdir,
+			.packtmp = packtmp,
+		};
 
-		ret = write_filtered_pack(&po_args,
-					  filter_to,
-					  find_pack_prefix(packdir, packtmp),
-					  &existing,
-					  &names);
+		if (!opts.destination)
+			opts.destination = packtmp;
+
+		ret = write_filtered_pack(&opts, &existing, &names);
 		if (ret)
 			goto cleanup;
 	}
diff --git a/repack.h b/repack.h
index 25a31ac0a0..6ef503f623 100644
--- a/repack.h
+++ b/repack.h
@@ -32,6 +32,14 @@ void pack_objects_args_release(struct pack_objects_args *args);
 void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 				  const char *base_name);
 
+struct write_pack_opts {
+	struct pack_objects_args *po_args;
+	const char *destination;
+	const char *pack_prefix;
+	const char *packdir;
+	const char *packtmp;
+};
+
 struct repository;
 struct packed_git;
 
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Oct 10, 2025, 12:28 UTC · re: Taylor Blau · lore

Re: [PATCH 40/49] builtin/repack.c: introduce `struct write_pack_opts`

On Sun, Sep 28, 2025 at 06:09:51PM -0400, Taylor Blau wrote:
Show 14 quoted lines
> diff --git a/builtin/repack.c b/builtin/repack.c
> index 5fed79e826..6df7c88085 100644
> --- a/builtin/repack.c
> +++ b/builtin/repack.c
> @@ -138,9 +138,7 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
>  	return finish_command(cmd);
>  }
>  
> -static int write_filtered_pack(const struct pack_objects_args *args,
> -			       const char *destination,
> -			       const char *pack_prefix,
> +static int write_filtered_pack(struct write_pack_opts *opts,
>  			       struct existing_packs *existing,
>  			       struct string_list *names)
Is there any reason why the new argument isn't marked as `const`?
Patrick
Taylor Blau· Oct 10, 2025, 23:13 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 40/49] builtin/repack.c: introduce `struct write_pack_opts`

On Fri, Oct 10, 2025 at 02:28:22PM +0200, Patrick Steinhardt wrote:
Show 17 quoted lines
> On Sun, Sep 28, 2025 at 06:09:51PM -0400, Taylor Blau wrote:
> > diff --git a/builtin/repack.c b/builtin/repack.c
> > index 5fed79e826..6df7c88085 100644
> > --- a/builtin/repack.c
> > +++ b/builtin/repack.c
> > @@ -138,9 +138,7 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
> >  	return finish_command(cmd);
> >  }
> >
> > -static int write_filtered_pack(const struct pack_objects_args *args,
> > -			       const char *destination,
> > -			       const char *pack_prefix,
> > +static int write_filtered_pack(struct write_pack_opts *opts,
> >  			       struct existing_packs *existing,
> >  			       struct string_list *names)
>
> Is there any reason why the new argument isn't marked as `const`?
None other than forgetfulness ;-).

I adjusted it locally, which makes me think that I've accumulated enough to merit sending out another round of this series. I'll plan on doing so sometime early next week once the review has settled down here.

Thanks, Taylor

Jeff King· Oct 15, 2025, 10:28 UTC · re: Taylor Blau · lore

Re: [PATCH 40/49] builtin/repack.c: introduce `struct write_pack_opts`

On Sun, Sep 28, 2025 at 06:09:51PM -0400, Taylor Blau wrote:
Show 5 quoted lines
> Instead of repeating those arguments for each function, let's extract an
> options struct called "write_pack_opts" which has these three parameters
> as member fields. While we're at it, add fields for "packdir," and
> "packtmp", both of which are static variables within the builtin, and
> need to be read from within these two functions.
Makes sense, although...
Show 7 quoted lines
> +		struct write_pack_opts opts = {
> +			.po_args = &po_args,
> +			.destination = filter_to,
> +			.pack_prefix = find_pack_prefix(packdir, packtmp),
> +			.packdir = packdir,
> +			.packtmp = packtmp,
> +		};

...since we are now passing packdir and packtmp anyway, and pack_prefix is derived from those, should the called function just do that derivation itself? Or do we expect that some callers may eventually use a different prefix?

Probably not a huge deal either way, but maybe an easy way to tighten up the interface a bit.

-Peff
Taylor Blau· Oct 15, 2025, 21:18 UTC · re: Jeff King · lore

Re: [PATCH 40/49] builtin/repack.c: introduce `struct write_pack_opts`

On Wed, Oct 15, 2025 at 06:28:33AM -0400, Jeff King wrote:
Show 25 quoted lines
> On Sun, Sep 28, 2025 at 06:09:51PM -0400, Taylor Blau wrote:
>
> > Instead of repeating those arguments for each function, let's extract an
> > options struct called "write_pack_opts" which has these three parameters
> > as member fields. While we're at it, add fields for "packdir," and
> > "packtmp", both of which are static variables within the builtin, and
> > need to be read from within these two functions.
>
> Makes sense, although...
>
> > +		struct write_pack_opts opts = {
> > +			.po_args = &po_args,
> > +			.destination = filter_to,
> > +			.pack_prefix = find_pack_prefix(packdir, packtmp),
> > +			.packdir = packdir,
> > +			.packtmp = packtmp,
> > +		};
>
> ...since we are now passing packdir and packtmp anyway, and pack_prefix
> is derived from those, should the called function just do that
> derivation itself? Or do we expect that some callers may eventually use
> a different prefix?
>
> Probably not a huge deal either way, but maybe an easy way to tighten up
> the interface a bit.

We do this a little further down in "repack: move `find_pack_prefix()` out of the builtin", which I tried to separate since I wanted to go incrementally here. But I should mention that that patch is upcoming to avoid readers wondering why we're not making the change sooner.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:09 UTC · re: Taylor Blau · lore

[PATCH 41/49] builtin/repack.c: use `write_pack_opts` within `write_cruft_pack()`

Similar to the changes made in the previous commit to `write_filtered_pack()`, teach `write_cruft_pack()` to take a `write_pack_opts` struct and use that where possible.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)
Show changes to builtin/repack.c +14 −13
diff --git a/builtin/repack.c b/builtin/repack.c
index 6df7c88085..501359c580 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -221,9 +221,7 @@ static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
 	strbuf_release(&buf);
 }
 
-static int write_cruft_pack(const struct pack_objects_args *args,
-			    const char *destination,
-			    const char *pack_prefix,
+static int write_cruft_pack(struct write_pack_opts *opts,
 			    const char *cruft_expiration,
 			    unsigned long combine_cruft_below_size,
 			    struct string_list *names,
@@ -234,9 +232,9 @@ static int write_cruft_pack(const struct pack_objects_args *args,
 	FILE *in;
 	int ret;
 	const char *scratch;
-	int local = skip_prefix(destination, packdir, &scratch);
+	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
 
-	prepare_pack_objects(&cmd, args, destination);
+	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
 
 	strvec_push(&cmd.args, "--cruft");
 	if (cruft_expiration)
@@ -267,7 +265,7 @@ static int write_cruft_pack(const struct pack_objects_args *args,
 	 */
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, names)
-		fprintf(in, "%s-%s.pack\n", pack_prefix, item->string);
+		fprintf(in, "%s-%s.pack\n", opts->pack_prefix, item->string);
 	if (combine_cruft_below_size && !cruft_expiration) {
 		combine_small_cruft_packs(in, combine_cruft_below_size,
 					  existing);
@@ -599,6 +597,13 @@ int cmd_repack(int argc,
 
 	if (pack_everything & PACK_CRUFT) {
 		const char *pack_prefix = find_pack_prefix(packdir, packtmp);
+		struct write_pack_opts opts = {
+			.po_args = &cruft_po_args,
+			.destination = packtmp,
+			.pack_prefix = pack_prefix,
+			.packtmp = packtmp,
+			.packdir = packdir,
+		};
 
 		if (!cruft_po_args.window)
 			cruft_po_args.window = xstrdup_or_null(po_args.window);
@@ -615,8 +620,7 @@ int cmd_repack(int argc,
 		cruft_po_args.quiet = po_args.quiet;
 		cruft_po_args.delta_base_offset = po_args.delta_base_offset;
 
-		ret = write_cruft_pack(&cruft_po_args, packtmp, pack_prefix,
-				       cruft_expiration,
+		ret = write_cruft_pack(&opts, cruft_expiration,
 				       combine_cruft_below_size, &names,
 				       &existing);
 		if (ret)
@@ -651,11 +655,8 @@ int cmd_repack(int argc,
 			 * pack, but rather removing all cruft packs from the
 			 * main repository regardless of size.
 			 */
-			ret = write_cruft_pack(&cruft_po_args, expire_to,
-					       pack_prefix,
-					       NULL,
-					       0ul,
-					       &names,
+			opts.destination = expire_to;
+			ret = write_cruft_pack(&opts, NULL, 0ul, &names,
 					       &existing);
 			if (ret)
 				goto cleanup;
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Oct 10, 2025, 12:28 UTC · re: Taylor Blau · lore

Re: [PATCH 41/49] builtin/repack.c: use `write_pack_opts` within `write_cruft_pack()`

On Sun, Sep 28, 2025 at 06:09:57PM -0400, Taylor Blau wrote:
Show 15 quoted lines
> diff --git a/builtin/repack.c b/builtin/repack.c
> index 6df7c88085..501359c580 100644
> --- a/builtin/repack.c
> +++ b/builtin/repack.c
> @@ -221,9 +221,7 @@ static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
>  	strbuf_release(&buf);
>  }
>  
> -static int write_cruft_pack(const struct pack_objects_args *args,
> -			    const char *destination,
> -			    const char *pack_prefix,
> +static int write_cruft_pack(struct write_pack_opts *opts,
>  			    const char *cruft_expiration,
>  			    unsigned long combine_cruft_below_size,
>  			    struct string_list *names,

Same question here regarding the constness. Subsequent patches also surface this question.

Patrick
Taylor Blau· Oct 10, 2025, 23:13 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 41/49] builtin/repack.c: use `write_pack_opts` within `write_cruft_pack()`

On Fri, Oct 10, 2025 at 02:28:38PM +0200, Patrick Steinhardt wrote:
Show 19 quoted lines
> On Sun, Sep 28, 2025 at 06:09:57PM -0400, Taylor Blau wrote:
> > diff --git a/builtin/repack.c b/builtin/repack.c
> > index 6df7c88085..501359c580 100644
> > --- a/builtin/repack.c
> > +++ b/builtin/repack.c
> > @@ -221,9 +221,7 @@ static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
> >  	strbuf_release(&buf);
> >  }
> >
> > -static int write_cruft_pack(const struct pack_objects_args *args,
> > -			    const char *destination,
> > -			    const char *pack_prefix,
> > +static int write_cruft_pack(struct write_pack_opts *opts,
> >  			    const char *cruft_expiration,
> >  			    unsigned long combine_cruft_below_size,
> >  			    struct string_list *names,
>
> Same question here regarding the constness. Subsequent patches also
> surface this question.

Yeah, there ended up being a little bit of fallout/churn after marking it as const earlier, but those are all fixed up in my local version of these later patches.

Thanks, Taylor

Jeff King· Oct 15, 2025, 10:29 UTC · re: Taylor Blau · lore

Re: [PATCH 41/49] builtin/repack.c: use `write_pack_opts` within `write_cruft_pack()`

On Sun, Sep 28, 2025 at 06:09:57PM -0400, Taylor Blau wrote:
Show 7 quoted lines
> -static int write_cruft_pack(const struct pack_objects_args *args,
> -			    const char *destination,
> -			    const char *pack_prefix,
> +static int write_cruft_pack(struct write_pack_opts *opts,
>  			    const char *cruft_expiration,
>  			    unsigned long combine_cruft_below_size,
>  			    struct string_list *names,

Following on the comment from the previous commit, I guess with my proposal we'd have to call find_pack_prefix() ourselves here in write_cruft_pack().

OTOH, this call...
Show 11 quoted lines
> @@ -599,6 +597,13 @@ int cmd_repack(int argc,
>  
>  	if (pack_everything & PACK_CRUFT) {
>  		const char *pack_prefix = find_pack_prefix(packdir, packtmp);
> +		struct write_pack_opts opts = {
> +			.po_args = &cruft_po_args,
> +			.destination = packtmp,
> +			.pack_prefix = pack_prefix,
> +			.packtmp = packtmp,
> +			.packdir = packdir,
> +		};
...would go away, so I think it is net the same number of lines.
-Peff
Taylor Blau· Sep 28, 2025, 22:10 UTC · re: Taylor Blau · lore

[PATCH 42/49] repack: move `find_pack_prefix()` out of the builtin

Both callers within the repack builtin which call functions that take a 'write_pack_opts' structure have the following pattern:

    struct write_pack_opts opts = {
        .packdir = packdir,
        .packtmp = packtmp,
        .pack_prefix = find_pack_prefix(packdir, packtmp),
        /* ... */
    };
    int ret = write_some_kind_of_pack(&opts, /* ... */);

, but both "packdir" and "packtmp" are fields within the write_pack_opts struct itself!

Instead of also computing the pack_prefix ahead of time, let's have the callees compute it themselves by moving `find_pack_prefix()` out of the repack builtin, and have it take a write_pack_opts pointer instead of the "packdir" and "packtmp" fields directly.

This avoids the callers having to do some prep work that is common between the two of them, but also avoids the potential pitfall of accidentally writing:

    .pack_prefix = find_pack_prefix(packtmp, packdir),
(which is well-typed) when the caller meant to instead write:
    .pack_prefix = find_pack_prefix(packdir, packtmp),
Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 20 ++++----------------
 repack.c         | 11 +++++++++++
 repack.h         |  3 ++-
 3 files changed, 17 insertions(+), 17 deletions(-)
Show changes to 3 files +17 −17

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 501359c580..4d86920618 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -149,6 +149,7 @@ static int write_filtered_pack(struct write_pack_opts *opts,
 	const char *caret;
 	const char *scratch;
 	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
+	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
 
@@ -173,7 +174,7 @@ static int write_filtered_pack(struct write_pack_opts *opts,
 	 */
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, names)
-		fprintf(in, "^%s-%s.pack\n", opts->pack_prefix, item->string);
+		fprintf(in, "^%s-%s.pack\n", pack_prefix, item->string);
 	for_each_string_list_item(item, &existing->non_kept_packs)
 		fprintf(in, "%s.pack\n", item->string);
 	for_each_string_list_item(item, &existing->cruft_packs)
@@ -233,6 +234,7 @@ static int write_cruft_pack(struct write_pack_opts *opts,
 	int ret;
 	const char *scratch;
 	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
+	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
 
@@ -265,7 +267,7 @@ static int write_cruft_pack(struct write_pack_opts *opts,
 	 */
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, names)
-		fprintf(in, "%s-%s.pack\n", opts->pack_prefix, item->string);
+		fprintf(in, "%s-%s.pack\n", pack_prefix, item->string);
 	if (combine_cruft_below_size && !cruft_expiration) {
 		combine_small_cruft_packs(in, combine_cruft_below_size,
 					  existing);
@@ -283,17 +285,6 @@ static int write_cruft_pack(struct write_pack_opts *opts,
 				       local);
 }
 
-static const char *find_pack_prefix(const char *packdir, const char *packtmp)
-{
-	const char *pack_prefix;
-	if (!skip_prefix(packtmp, packdir, &pack_prefix))
-		die(_("pack prefix %s does not begin with objdir %s"),
-		    packtmp, packdir);
-	if (*pack_prefix == '/')
-		pack_prefix++;
-	return pack_prefix;
-}
-
 int cmd_repack(int argc,
 	       const char **argv,
 	       const char *prefix,
@@ -596,11 +587,9 @@ int cmd_repack(int argc,
 	}
 
 	if (pack_everything & PACK_CRUFT) {
-		const char *pack_prefix = find_pack_prefix(packdir, packtmp);
 		struct write_pack_opts opts = {
 			.po_args = &cruft_po_args,
 			.destination = packtmp,
-			.pack_prefix = pack_prefix,
 			.packtmp = packtmp,
 			.packdir = packdir,
 		};
@@ -667,7 +656,6 @@ int cmd_repack(int argc,
 		struct write_pack_opts opts = {
 			.po_args = &po_args,
 			.destination = filter_to,
-			.pack_prefix = find_pack_prefix(packdir, packtmp),
 			.packdir = packdir,
 			.packtmp = packtmp,
 		};
diff --git a/repack.c b/repack.c
index 1d485e0112..c4326a532d 100644
--- a/repack.c
+++ b/repack.c
@@ -66,6 +66,17 @@ void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 	strbuf_release(&buf);
 }
 
+const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts)
+{
+	const char *pack_prefix;
+	if (!skip_prefix(opts->packtmp, opts->packdir, &pack_prefix))
+		die(_("pack prefix %s does not begin with objdir %s"),
+		    opts->packtmp, opts->packdir);
+	if (*pack_prefix == '/')
+		pack_prefix++;
+	return pack_prefix;
+}
+
 #define DELETE_PACK 1
 #define RETAIN_PACK 2
 
diff --git a/repack.h b/repack.h
index 6ef503f623..46d2312fa9 100644
--- a/repack.h
+++ b/repack.h
@@ -35,11 +35,12 @@ void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 struct write_pack_opts {
 	struct pack_objects_args *po_args;
 	const char *destination;
-	const char *pack_prefix;
 	const char *packdir;
 	const char *packtmp;
 };
 
+const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts);
+
 struct repository;
 struct packed_git;
 
-- 
2.51.0.243.g16eca91f2c0
Jeff King· Oct 15, 2025, 10:32 UTC · re: Taylor Blau · lore

Re: [PATCH 42/49] repack: move `find_pack_prefix()` out of the builtin

On Sun, Sep 28, 2025 at 06:10:02PM -0400, Taylor Blau wrote:
Show 18 quoted lines
> Both callers within the repack builtin which call functions that take a
> 'write_pack_opts' structure have the following pattern:
> 
>     struct write_pack_opts opts = {
>         .packdir = packdir,
>         .packtmp = packtmp,
>         .pack_prefix = find_pack_prefix(packdir, packtmp),
>         /* ... */
>     };
>     int ret = write_some_kind_of_pack(&opts, /* ... */);
> 
> , but both "packdir" and "packtmp" are fields within the write_pack_opts
> struct itself!
> 
> Instead of also computing the pack_prefix ahead of time, let's have the
> callees compute it themselves by moving `find_pack_prefix()` out of the
> repack builtin, and have it take a write_pack_opts pointer instead of
> the "packdir" and "packtmp" fields directly.

Oh. You'd really think by now that I'd have learned to read ahead in the series before commenting.

So you can ignore my responses to the previous two patches. This does what I propose, plus takes it further by using the opts struct itself.

-Peff
Taylor Blau· Oct 15, 2025, 21:19 UTC · re: Jeff King · lore

Re: [PATCH 42/49] repack: move `find_pack_prefix()` out of the builtin

On Wed, Oct 15, 2025 at 06:32:20AM -0400, Jeff King wrote:
Show 26 quoted lines
> On Sun, Sep 28, 2025 at 06:10:02PM -0400, Taylor Blau wrote:
>
> > Both callers within the repack builtin which call functions that take a
> > 'write_pack_opts' structure have the following pattern:
> >
> >     struct write_pack_opts opts = {
> >         .packdir = packdir,
> >         .packtmp = packtmp,
> >         .pack_prefix = find_pack_prefix(packdir, packtmp),
> >         /* ... */
> >     };
> >     int ret = write_some_kind_of_pack(&opts, /* ... */);
> >
> > , but both "packdir" and "packtmp" are fields within the write_pack_opts
> > struct itself!
> >
> > Instead of also computing the pack_prefix ahead of time, let's have the
> > callees compute it themselves by moving `find_pack_prefix()` out of the
> > repack builtin, and have it take a write_pack_opts pointer instead of
> > the "packdir" and "packtmp" fields directly.
>
> Oh. You'd really think by now that I'd have learned to read ahead in the
> series before commenting.
>
> So you can ignore my responses to the previous two patches. This does
> what I propose, plus takes it further by using the opts struct itself.
;-).

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:10 UTC · re: Taylor Blau · lore

[PATCH 43/49] repack: extract `write_pack_opts_is_local()`

Similar to the previous commit, the functions `write_cruft_pack()` and `write_filtered_pack()` both compute a "local" variable via the exact same mechanism:

    const char *scratch;
    int local = skip_prefix(opts->destination, opts->packdir, &scratch);

Not only does this cause us to repeat the same pair of lines, it also introduces an unnecessary "scratch" variable that is common between both functions.

Instead of repeating ourselves, let's extract that functionality into a new function in the repack.h API called "write_pack_opts_is_local()". That function takes a pointer to a "struct write_pack_opts" (which has as fields both "destination" and "packdir"), and can encapsulate the dangling "scratch" field.

Extract that function and make it visible within the repack.h API, and use it within both `write_cruft_pack()` and `write_filtered_pack()`.

The remaining duplication (that is, that both `write_cruft_pack()` and `write_filtered_pack()` still both call `write_pack_opts_is_local()`) will be addressed in the following commit.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 6 ++----
 repack.c         | 6 ++++++
 repack.h         | 1 +
 3 files changed, 9 insertions(+), 4 deletions(-)
Show changes to 3 files +9 −4

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 4d86920618..be8e6689fc 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -147,8 +147,7 @@ static int write_filtered_pack(struct write_pack_opts *opts,
 	FILE *in;
 	int ret;
 	const char *caret;
-	const char *scratch;
-	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
+	int local = write_pack_opts_is_local(opts);
 	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
@@ -232,8 +231,7 @@ static int write_cruft_pack(struct write_pack_opts *opts,
 	struct string_list_item *item;
 	FILE *in;
 	int ret;
-	const char *scratch;
-	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
+	int local = write_pack_opts_is_local(opts);
 	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
diff --git a/repack.c b/repack.c
index c4326a532d..7af297ae48 100644
--- a/repack.c
+++ b/repack.c
@@ -77,6 +77,12 @@ const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts)
 	return pack_prefix;
 }
 
+int write_pack_opts_is_local(struct write_pack_opts *opts)
+{
+	const char *scratch;
+	return skip_prefix(opts->destination, opts->packdir, &scratch);
+}
+
 #define DELETE_PACK 1
 #define RETAIN_PACK 2
 
diff --git a/repack.h b/repack.h
index 46d2312fa9..16f2de2ea9 100644
--- a/repack.h
+++ b/repack.h
@@ -40,6 +40,7 @@ struct write_pack_opts {
 };
 
 const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts);
+int write_pack_opts_is_local(struct write_pack_opts *opts);
 
 struct repository;
 struct packed_git;
-- 
2.51.0.243.g16eca91f2c0
Patrick Steinhardt· Oct 10, 2025, 12:28 UTC · re: Taylor Blau · lore

Re: [PATCH 43/49] repack: extract `write_pack_opts_is_local()`

On Sun, Sep 28, 2025 at 06:10:08PM -0400, Taylor Blau wrote:
Show 17 quoted lines
> diff --git a/repack.c b/repack.c
> index c4326a532d..7af297ae48 100644
> --- a/repack.c
> +++ b/repack.c
> @@ -77,6 +77,12 @@ const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts)
>  	return pack_prefix;
>  }
>  
> +int write_pack_opts_is_local(struct write_pack_opts *opts)
> +{
> +	const char *scratch;
> +	return skip_prefix(opts->destination, opts->packdir, &scratch);
> +}
> +
>  #define DELETE_PACK 1
>  #define RETAIN_PACK 2
>  

Now that we accept booleans in our codebase, should this function return a `bool` instead?

Show 9 quoted lines
> diff --git a/repack.h b/repack.h
> index 46d2312fa9..16f2de2ea9 100644
> --- a/repack.h
> +++ b/repack.h
> @@ -40,6 +40,7 @@ struct write_pack_opts {
>  };
>  
>  const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts);
> +int write_pack_opts_is_local(struct write_pack_opts *opts);

It would be good if all of these these new library functions could come with some documentation what they actually do.

Patrick
Taylor Blau· Oct 15, 2025, 21:21 UTC · re: Patrick Steinhardt · lore

Re: [PATCH 43/49] repack: extract `write_pack_opts_is_local()`

On Fri, Oct 10, 2025 at 02:28:57PM +0200, Patrick Steinhardt wrote:
Show 21 quoted lines
> On Sun, Sep 28, 2025 at 06:10:08PM -0400, Taylor Blau wrote:
> > diff --git a/repack.c b/repack.c
> > index c4326a532d..7af297ae48 100644
> > --- a/repack.c
> > +++ b/repack.c
> > @@ -77,6 +77,12 @@ const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts)
> >  	return pack_prefix;
> >  }
> >
> > +int write_pack_opts_is_local(struct write_pack_opts *opts)
> > +{
> > +	const char *scratch;
> > +	return skip_prefix(opts->destination, opts->packdir, &scratch);
> > +}
> > +
> >  #define DELETE_PACK 1
> >  #define RETAIN_PACK 2
> >
>
> Now that we accept booleans in our codebase, should this function return
> a `bool` instead?

Yeah, I agree that it should eventually return "bool", but I don't want to make the change in this patch, at least within the context of this series where I'm trying to keep the changes-per-patch as minimal as possible.

Show 12 quoted lines
> > diff --git a/repack.h b/repack.h
> > index 46d2312fa9..16f2de2ea9 100644
> > --- a/repack.h
> > +++ b/repack.h
> > @@ -40,6 +40,7 @@ struct write_pack_opts {
> >  };
> >
> >  const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts);
> > +int write_pack_opts_is_local(struct write_pack_opts *opts);
>
> It would be good if all of these these new library functions could come
> with some documentation what they actually do.

Fully agreed. I think there is enough here that I'd rather pursue this as a separate step, but at least this series moves us in a good direction to document things by putting the functions in a clean header that is more amenable to documentation than the builtin is.

Thanks, Taylor

Jeff King· Oct 15, 2025, 10:35 UTC · re: Taylor Blau · lore

Re: [PATCH 43/49] repack: extract `write_pack_opts_is_local()`

On Sun, Sep 28, 2025 at 06:10:08PM -0400, Taylor Blau wrote:
Show 10 quoted lines
> Similar to the previous commit, the functions `write_cruft_pack()` and
> `write_filtered_pack()` both compute a "local" variable via the exact
> same mechanism:
> 
>     const char *scratch;
>     int local = skip_prefix(opts->destination, opts->packdir, &scratch);
> 
> Not only does this cause us to repeat the same pair of lines, it also
> introduces an unnecessary "scratch" variable that is common between both
> functions.

Hmm. If we are not looking at "scratch", then does that mean skip_prefix() is the wrong function to use? I.e., should this just be:

  int local = starts_with(opts->destinations, opts->packdir);
?

It may still be worth pulling into a separate function to encapsulate the policy logic, but that function could be using starts_with() itself.

-Peff
Taylor Blau· Oct 15, 2025, 21:25 UTC · re: Jeff King · lore

Re: [PATCH 43/49] repack: extract `write_pack_opts_is_local()`

On Wed, Oct 15, 2025 at 06:35:16AM -0400, Jeff King wrote:
Show 22 quoted lines
> On Sun, Sep 28, 2025 at 06:10:08PM -0400, Taylor Blau wrote:
>
> > Similar to the previous commit, the functions `write_cruft_pack()` and
> > `write_filtered_pack()` both compute a "local" variable via the exact
> > same mechanism:
> >
> >     const char *scratch;
> >     int local = skip_prefix(opts->destination, opts->packdir, &scratch);
> >
> > Not only does this cause us to repeat the same pair of lines, it also
> > introduces an unnecessary "scratch" variable that is common between both
> > functions.
>
> Hmm. If we are not looking at "scratch", then does that mean
> skip_prefix() is the wrong function to use? I.e., should this just be:
>
>   int local = starts_with(opts->destinations, opts->packdir);
>
> ?
>
> It may still be worth pulling into a separate function to encapsulate
> the policy logic, but that function could be using starts_with() itself.

When responding to Patrick's message on this patch I was inclined to not make any more changes in this patch than strictly necessary. But having two suggestions which are both easy to make, and (more importantly) easy to verify the correctness of pushed me to go ahead and make both changes.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:10 UTC · re: Taylor Blau · lore

[PATCH 44/49] builtin/repack.c: pass `write_pack_opts` to `finish_pack_objects_cmd()`

To prepare to move the `finish_pack_objects_cmd()` function out of the builtin and into the repack.h API, there are a couple of things we need to do first:

 - First, let's take advantage of `write_pack_opts_is_local()` function
   introduced in the previous commit instead of passing "local"
   explicitly.
 - Let's also avoid referring to the static 'packtmp' field within
   builtin/repack.c by instead accessing it through the write_pack_opts
   argument.

There are three callers which need to adjust themselves in order to account for this change. The callers which reside in write_cruft_pack() and write_filtered_pack() both already have an "opts" in scope, so they can pass it through transparently.

The other call (at the bottom of `cmd_repack()`) needs to initialize its own write_pack_opts to pass the necessary fields over to the direct call to `finish_pack_objects_cmd()`.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 24 ++++++++++++++----------
 1 file changed, 14 insertions(+), 10 deletions(-)
Show changes to builtin/repack.c +14 −10
diff --git a/builtin/repack.c b/builtin/repack.c
index be8e6689fc..8db95305c8 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -108,11 +108,12 @@ static int repack_config(const char *var, const char *value,
 }
 
 static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
+				   struct write_pack_opts *opts,
 				   struct child_process *cmd,
-				   struct string_list *names,
-				   int local)
+				   struct string_list *names)
 {
 	FILE *out;
+	int local = write_pack_opts_is_local(opts);
 	struct strbuf line = STRBUF_INIT;
 
 	out = xfdopen(cmd->out, "r");
@@ -128,7 +129,8 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 		 */
 		if (local) {
 			item = string_list_append(names, line.buf);
-			item->util = generated_pack_populate(line.buf, packtmp);
+			item->util = generated_pack_populate(line.buf,
+							     opts->packtmp);
 		}
 	}
 	fclose(out);
@@ -147,7 +149,6 @@ static int write_filtered_pack(struct write_pack_opts *opts,
 	FILE *in;
 	int ret;
 	const char *caret;
-	int local = write_pack_opts_is_local(opts);
 	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
@@ -183,8 +184,8 @@ static int write_filtered_pack(struct write_pack_opts *opts,
 		fprintf(in, "%s%s.pack\n", caret, item->string);
 	fclose(in);
 
-	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
-				       local);
+	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
+				       names);
 }
 
 static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
@@ -231,7 +232,6 @@ static int write_cruft_pack(struct write_pack_opts *opts,
 	struct string_list_item *item;
 	FILE *in;
 	int ret;
-	int local = write_pack_opts_is_local(opts);
 	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
@@ -279,8 +279,8 @@ static int write_cruft_pack(struct write_pack_opts *opts,
 		fprintf(in, "%s.pack\n", item->string);
 	fclose(in);
 
-	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
-				       local);
+	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
+				       names);
 }
 
 int cmd_repack(int argc,
@@ -294,6 +294,7 @@ int cmd_repack(int argc,
 	struct existing_packs existing = EXISTING_PACKS_INIT;
 	struct pack_geometry geometry = { 0 };
 	struct tempfile *refs_snapshot = NULL;
+	struct write_pack_opts opts = { 0 };
 	int i, ret;
 	int show_progress;
 
@@ -560,7 +561,10 @@ int cmd_repack(int argc,
 		fclose(in);
 	}
 
-	ret = finish_pack_objects_cmd(repo->hash_algo, &cmd, &names, 1);
+	opts.packdir = packdir;
+	opts.destination = packdir;
+	opts.packtmp = packtmp;
+	ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
 	if (ret)
 		goto cleanup;
 
-- 
2.51.0.243.g16eca91f2c0
Jeff King· Oct 15, 2025, 11:01 UTC · re: Taylor Blau · lore

Re: [PATCH 44/49] builtin/repack.c: pass `write_pack_opts` to `finish_pack_objects_cmd()`

On Sun, Sep 28, 2025 at 06:10:14PM -0400, Taylor Blau wrote:
Show 20 quoted lines
>  int cmd_repack(int argc,
> @@ -294,6 +294,7 @@ int cmd_repack(int argc,
>  	struct existing_packs existing = EXISTING_PACKS_INIT;
>  	struct pack_geometry geometry = { 0 };
>  	struct tempfile *refs_snapshot = NULL;
> +	struct write_pack_opts opts = { 0 };
>  	int i, ret;
>  	int show_progress;
>  
> @@ -560,7 +561,10 @@ int cmd_repack(int argc,
>  		fclose(in);
>  	}
>  
> -	ret = finish_pack_objects_cmd(repo->hash_algo, &cmd, &names, 1);
> +	opts.packdir = packdir;
> +	opts.destination = packdir;
> +	opts.packtmp = packtmp;
> +	ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
>  	if (ret)
>  		goto cleanup;

This looks a bit funny: we are creating a write_pack_opts struct but only filling in some of the fields. Notably po_args is NULL. It's zero'd, which is good, but as a matter of API design, how can a caller know which fields are important for a given function and which are not?

I guess I could have expected cmd_repack() to make a fully fleshed-out write_pack_opts up front, and then pass it to all functions which need it. Rather than making a custom one for this one function.

But I guess that gets a little weird, because it really has at least _two_ such opts: one for the regular pack-objects invocation and one for the cruft pack. And really maybe one more, for the filtered pack? And the conditional blocks for cruft and filter_options packs shadow the "opts" variable. Yuck.

I'm not sure if there's a good solution. My inclination is to say that po_args should not be in "opts" at all, and that each function should take a separate po_args as appropriate. And then we could just have one write_pack_opts in cmd_repack(). But there is another snag: we sometimes override the "destination" field with "filter_to" or "expire_to".

So the struct really isn't "here are the opts for repacking", but rather "let's marshal some arguments into a struct for one function call". Which of course is a bit verbose in C, hence these extra shadowed "opts" variables.

So I dunno that I am really asking for any change. This might be the least bad way of writing it. I guess if we could avoid the shadowing like this:

Show changes to builtin/repack.c +16 −12
diff --git a/builtin/repack.c b/builtin/repack.c
index ad60c4290d..4e4519b180 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -108,7 +108,6 @@ int cmd_repack(int argc,
 	struct existing_packs existing = EXISTING_PACKS_INIT;
 	struct pack_geometry geometry = { 0 };
 	struct tempfile *refs_snapshot = NULL;
-	struct write_pack_opts opts = { 0 };
 	int i, ret;
 	int show_progress;
 
@@ -372,12 +371,15 @@ int cmd_repack(int argc,
 		fclose(in);
 	}
 
-	opts.packdir = packdir;
-	opts.destination = packdir;
-	opts.packtmp = packtmp;
-	ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
-	if (ret)
-		goto cleanup;
+	{
+		struct write_pack_opts opts = { 0 };
+		opts.packdir = packdir;
+		opts.destination = packdir;
+		opts.packtmp = packtmp;
+		ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
+		if (ret)
+			goto cleanup;
+	}
 
 	if (!names.nr) {
 		if (!po_args.quiet)

That might be nice, but introducing the extra block is kind of ugly. Of
course you can get really fancy with a compound literal, like:

@@ -372,10 +371,13 @@ int cmd_repack(int argc,
 		fclose(in);
 	}
 
-	opts.packdir = packdir;
-	opts.destination = packdir;
-	opts.packtmp = packtmp;
-	ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
+	ret = finish_pack_objects_cmd(repo->hash_algo,
+				      &(struct write_pack_opts) {
+					.packdir = packdir,
+					.destination = packdir,
+					.packtmp = packtmp,
+				      },
+				      &cmd, &names);
 	if (ret)
 		goto cleanup;
 

But that may be a bridge too far.

-Peff
Taylor Blau· Oct 15, 2025, 21:35 UTC · re: Jeff King · lore

Re: [PATCH 44/49] builtin/repack.c: pass `write_pack_opts` to `finish_pack_objects_cmd()`

On Wed, Oct 15, 2025 at 07:01:55AM -0400, Jeff King wrote:
Show 37 quoted lines
> On Sun, Sep 28, 2025 at 06:10:14PM -0400, Taylor Blau wrote:
>
> >  int cmd_repack(int argc,
> > @@ -294,6 +294,7 @@ int cmd_repack(int argc,
> >  	struct existing_packs existing = EXISTING_PACKS_INIT;
> >  	struct pack_geometry geometry = { 0 };
> >  	struct tempfile *refs_snapshot = NULL;
> > +	struct write_pack_opts opts = { 0 };
> >  	int i, ret;
> >  	int show_progress;
> >
> > @@ -560,7 +561,10 @@ int cmd_repack(int argc,
> >  		fclose(in);
> >  	}
> >
> > -	ret = finish_pack_objects_cmd(repo->hash_algo, &cmd, &names, 1);
> > +	opts.packdir = packdir;
> > +	opts.destination = packdir;
> > +	opts.packtmp = packtmp;
> > +	ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
> >  	if (ret)
> >  		goto cleanup;
>
> This looks a bit funny: we are creating a write_pack_opts struct but
> only filling in some of the fields. Notably po_args is NULL. It's
> zero'd, which is good, but as a matter of API design, how can a caller
> know which fields are important for a given function and which are not?
>
> I guess I could have expected cmd_repack() to make a fully fleshed-out
> write_pack_opts up front, and then pass it to all functions which need
> it. Rather than making a custom one for this one function.
>
> But I guess that gets a little weird, because it really has at least
> _two_ such opts: one for the regular pack-objects invocation and one for
> the cruft pack. And really maybe one more, for the filtered pack? And
> the conditional blocks for cruft and filter_options packs shadow the
> "opts" variable. Yuck.

Yeah, it is unfortunate that not all instances where we generate a pack use the same pack destination, or even the same set of options to pack-objects.

Show 10 quoted lines
> I'm not sure if there's a good solution. My inclination is to say that
> po_args should not be in "opts" at all, and that each function should
> take a separate po_args as appropriate. And then we could just have one
> write_pack_opts in cmd_repack(). But there is another snag: we sometimes
> override the "destination" field with "filter_to" or "expire_to".
>
> So the struct really isn't "here are the opts for repacking", but rather
> "let's marshal some arguments into a struct for one function call".
> Which of course is a bit verbose in C, hence these extra shadowed "opts"
> variables.

I think that only solves part of the problem, though. Among the other spots that call functions which take a `struct write_pack_opts`, not all of them use the same destination, so we'd still have to create multiple `struct write_pack_opts` instances in `cmd_repack()`, or pass the destination separately, which feels equally odd to me.

Show 38 quoted lines
> So I dunno that I am really asking for any change. This might be the
> least bad way of writing it. I guess if we could avoid the shadowing
> like this:
>
> diff --git a/builtin/repack.c b/builtin/repack.c
> index ad60c4290d..4e4519b180 100644
> --- a/builtin/repack.c
> +++ b/builtin/repack.c
> @@ -108,7 +108,6 @@ int cmd_repack(int argc,
>  	struct existing_packs existing = EXISTING_PACKS_INIT;
>  	struct pack_geometry geometry = { 0 };
>  	struct tempfile *refs_snapshot = NULL;
> -	struct write_pack_opts opts = { 0 };
>  	int i, ret;
>  	int show_progress;
>
> @@ -372,12 +371,15 @@ int cmd_repack(int argc,
>  		fclose(in);
>  	}
>
> -	opts.packdir = packdir;
> -	opts.destination = packdir;
> -	opts.packtmp = packtmp;
> -	ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
> -	if (ret)
> -		goto cleanup;
> +	{
> +		struct write_pack_opts opts = { 0 };
> +		opts.packdir = packdir;
> +		opts.destination = packdir;
> +		opts.packtmp = packtmp;
> +		ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
> +		if (ret)
> +			goto cleanup;
> +	}
>
>  	if (!names.nr) {
>  		if (!po_args.quiet)

Yeah, I think that is the most reasonable outcome, at least within the scope of this patch/series.

Show 23 quoted lines
> That might be nice, but introducing the extra block is kind of ugly. Of
> course you can get really fancy with a compound literal, like:
>
> @@ -372,10 +371,13 @@ int cmd_repack(int argc,
>  		fclose(in);
>  	}
>
> -	opts.packdir = packdir;
> -	opts.destination = packdir;
> -	opts.packtmp = packtmp;
> -	ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
> +	ret = finish_pack_objects_cmd(repo->hash_algo,
> +				      &(struct write_pack_opts) {
> +					.packdir = packdir,
> +					.destination = packdir,
> +					.packtmp = packtmp,
> +				      },
> +				      &cmd, &names);
>  	if (ret)
>  		goto cleanup;
>
>
> But that may be a bridge too far.

As much as I'd like to use a compound literal here, 9b2527caa4 (CodingGuidelines: document test balloons in flight, 2025-07-23) prevents us from adding new instances of them in our codebase for the time being.

Thanks, Taylor

Taylor Blau· Sep 28, 2025, 22:10 UTC · re: Taylor Blau · lore

[PATCH 45/49] repack: move `finish_pack_objects_cmd()` out of the builtin

In a similar spirit as the previous commit(s), now that the function `finish_pack_objects_cmd()` has no explicit dependencies within the repack builtin, let's extract it.

This prepares us to extract the remaining two functions within the repack builtin that explicitly write packfiles, which are `write_cruft_pack()` and `write_filtered_pack()`, which will be done in the future commits.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 33 ---------------------------------
 repack.c         | 33 +++++++++++++++++++++++++++++++++
 repack.h         |  5 +++++
 3 files changed, 38 insertions(+), 33 deletions(-)
Show changes to 3 files +38 −33

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 8db95305c8..836a006607 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,39 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
-				   struct write_pack_opts *opts,
-				   struct child_process *cmd,
-				   struct string_list *names)
-{
-	FILE *out;
-	int local = write_pack_opts_is_local(opts);
-	struct strbuf line = STRBUF_INIT;
-
-	out = xfdopen(cmd->out, "r");
-	while (strbuf_getline_lf(&line, out) != EOF) {
-		struct string_list_item *item;
-
-		if (line.len != algop->hexsz)
-			die(_("repack: Expecting full hex object ID lines only "
-			      "from pack-objects."));
-		/*
-		 * Avoid putting packs written outside of the repository in the
-		 * list of names.
-		 */
-		if (local) {
-			item = string_list_append(names, line.buf);
-			item->util = generated_pack_populate(line.buf,
-							     opts->packtmp);
-		}
-	}
-	fclose(out);
-
-	strbuf_release(&line);
-
-	return finish_command(cmd);
-}
-
 static int write_filtered_pack(struct write_pack_opts *opts,
 			       struct existing_packs *existing,
 			       struct string_list *names)
diff --git a/repack.c b/repack.c
index 7af297ae48..8a0e4789fa 100644
--- a/repack.c
+++ b/repack.c
@@ -83,6 +83,39 @@ int write_pack_opts_is_local(struct write_pack_opts *opts)
 	return skip_prefix(opts->destination, opts->packdir, &scratch);
 }
 
+int finish_pack_objects_cmd(const struct git_hash_algo *algop,
+			    struct write_pack_opts *opts,
+			    struct child_process *cmd,
+			    struct string_list *names)
+{
+	FILE *out;
+	int local = write_pack_opts_is_local(opts);
+	struct strbuf line = STRBUF_INIT;
+
+	out = xfdopen(cmd->out, "r");
+	while (strbuf_getline_lf(&line, out) != EOF) {
+		struct string_list_item *item;
+
+		if (line.len != algop->hexsz)
+			die(_("repack: Expecting full hex object ID lines only "
+			      "from pack-objects."));
+		/*
+		 * Avoid putting packs written outside of the repository in the
+		 * list of names.
+		 */
+		if (local) {
+			item = string_list_append(names, line.buf);
+			item->util = generated_pack_populate(line.buf,
+							     opts->packtmp);
+		}
+	}
+	fclose(out);
+
+	strbuf_release(&line);
+
+	return finish_command(cmd);
+}
+
 #define DELETE_PACK 1
 #define RETAIN_PACK 2
 
diff --git a/repack.h b/repack.h
index 16f2de2ea9..9351293233 100644
--- a/repack.h
+++ b/repack.h
@@ -42,6 +42,11 @@ struct write_pack_opts {
 const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts);
 int write_pack_opts_is_local(struct write_pack_opts *opts);
 
+int finish_pack_objects_cmd(const struct git_hash_algo *algop,
+			    struct write_pack_opts *opts,
+			    struct child_process *cmd,
+			    struct string_list *names);
+
 struct repository;
 struct packed_git;
 
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:10 UTC · re: Taylor Blau · lore

[PATCH 46/49] repack: move `pack_kept_objects` to `struct pack_objects_args`

The "pack_kept_objects" variable is defined as static to the repack builtin, but is inherently related to the pack-objects arguments that the builtin uses when generating new packs.

Move that field into the "struct pack_objects_args", and shuffle around where we append the corresponding command-line option when preparing a pack-objects process. Specifically:

 - `write_cruft_pack()` always wants to pass "--honor-pack-keep", so
   explicitly set the `pack_kept_objects` field to "0" when initializing
   the `write_pack_opts` struct before calling `write_cruft_pack()`.
 - `write_filtered_pack()` no longer needs to handle writing the
   command-line option "--honor-pack-keep" when preparing a pack-objects
   process, since its call to `prepare_pack_objects()` will have already
   taken care of that.
   `write_filtered_pack()` also reads the `pack_kept_objects` field to
   determine whether to write the existing kept packs with a leading "^"
   character, so update that to read through the `po_args` pointer
   instead.
 - `cmd_repack()` also no longer has to write the "--honor-pack-keep"
   flag explicitly, since this is also handled via its call to
   `prepare_pack_objects()`.

Since there is a default value for "pack_kept_objects" that relies on whether or not we are writing a bitmap (and not writing a MIDX), extract a default initializer for `struct pack_objects_args` that keeps this conditional default behavior.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c  | 20 +++++++-------------
 repack-geometry.c |  5 ++---
 repack.c          |  2 ++
 repack.h          |  9 ++++++---
 4 files changed, 17 insertions(+), 19 deletions(-)
Show changes to 4 files +17 −19

builtin/repack.c, repack-geometry.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 836a006607..9d89217b77 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -33,7 +33,6 @@
 #define RETAIN_PACK 2
 
 static int pack_everything;
-static int pack_kept_objects = -1;
 static int write_bitmaps = -1;
 static int use_delta_islands;
 static int run_update_server_info = 1;
@@ -68,7 +67,7 @@ static int repack_config(const char *var, const char *value,
 		return 0;
 	}
 	if (!strcmp(var, "repack.packkeptobjects")) {
-		pack_kept_objects = git_config_bool(var, value);
+		po_args->pack_kept_objects = git_config_bool(var, value);
 		return 0;
 	}
 	if (!strcmp(var, "repack.writebitmaps") ||
@@ -122,8 +121,6 @@ static int write_filtered_pack(struct write_pack_opts *opts,
 
 	strvec_push(&cmd.args, "--stdin-packs");
 
-	if (!pack_kept_objects)
-		strvec_push(&cmd.args, "--honor-pack-keep");
 	for_each_string_list_item(item, &existing->kept_packs)
 		strvec_pushf(&cmd.args, "--keep-pack=%s", item->string);
 
@@ -146,7 +143,7 @@ static int write_filtered_pack(struct write_pack_opts *opts,
 		fprintf(in, "%s.pack\n", item->string);
 	for_each_string_list_item(item, &existing->cruft_packs)
 		fprintf(in, "%s.pack\n", item->string);
-	caret = pack_kept_objects ? "" : "^";
+	caret = opts->po_args->pack_kept_objects ? "" : "^";
 	for_each_string_list_item(item, &existing->kept_packs)
 		fprintf(in, "%s%s.pack\n", caret, item->string);
 	fclose(in);
@@ -208,7 +205,6 @@ static int write_cruft_pack(struct write_pack_opts *opts,
 		strvec_pushf(&cmd.args, "--cruft-expiration=%s",
 			     cruft_expiration);
 
-	strvec_push(&cmd.args, "--honor-pack-keep");
 	strvec_push(&cmd.args, "--non-empty");
 
 	cmd.in = -1;
@@ -333,7 +329,7 @@ int cmd_repack(int argc,
 		OPT_UNSIGNED(0, "max-pack-size", &po_args.max_pack_size,
 			     N_("maximum size of each packfile")),
 		OPT_PARSE_LIST_OBJECTS_FILTER(&po_args.filter_options),
-		OPT_BOOL(0, "pack-kept-objects", &pack_kept_objects,
+		OPT_BOOL(0, "pack-kept-objects", &po_args.pack_kept_objects,
 				N_("repack objects in packs marked with .keep")),
 		OPT_STRING_LIST(0, "keep-pack", &keep_pack_list, N_("name"),
 				N_("do not repack this pack")),
@@ -379,8 +375,8 @@ int cmd_repack(int argc,
 		    (!(pack_everything & ALL_INTO_ONE) || !is_bare_repository()))
 			write_bitmaps = 0;
 	}
-	if (pack_kept_objects < 0)
-		pack_kept_objects = write_bitmaps > 0 && !write_midx;
+	if (po_args.pack_kept_objects < 0)
+		po_args.pack_kept_objects = write_bitmaps > 0 && !write_midx;
 
 	if (write_bitmaps && !(pack_everything & ALL_INTO_ONE) && !write_midx)
 		die(_(incremental_bitmap_conflict_error));
@@ -421,8 +417,7 @@ int cmd_repack(int argc,
 	if (geometry.split_factor) {
 		if (pack_everything)
 			die(_("options '%s' and '%s' cannot be used together"), "--geometric", "-A/-a");
-		pack_geometry_init(&geometry, &existing, &po_args,
-				   pack_kept_objects);
+		pack_geometry_init(&geometry, &existing, &po_args);
 		pack_geometry_split(&geometry);
 	}
 
@@ -431,8 +426,6 @@ int cmd_repack(int argc,
 	show_progress = !po_args.quiet && isatty(2);
 
 	strvec_push(&cmd.args, "--keep-true-parents");
-	if (!pack_kept_objects)
-		strvec_push(&cmd.args, "--honor-pack-keep");
 	for (i = 0; i < keep_pack_list.nr; i++)
 		strvec_pushf(&cmd.args, "--keep-pack=%s",
 			     keep_pack_list.items[i].string);
@@ -577,6 +570,7 @@ int cmd_repack(int argc,
 		cruft_po_args.local = po_args.local;
 		cruft_po_args.quiet = po_args.quiet;
 		cruft_po_args.delta_base_offset = po_args.delta_base_offset;
+		cruft_po_args.pack_kept_objects = 0;
 
 		ret = write_cruft_pack(&opts, cruft_expiration,
 				       combine_cruft_below_size, &names,
diff --git a/repack-geometry.c b/repack-geometry.c
index a879f2fe49..fd4a89a115 100644
--- a/repack-geometry.c
+++ b/repack-geometry.c
@@ -26,8 +26,7 @@ static int pack_geometry_cmp(const void *va, const void *vb)
 
 void pack_geometry_init(struct pack_geometry *geometry,
 			struct existing_packs *existing,
-			const struct pack_objects_args *args,
-			int pack_kept_objects)
+			const struct pack_objects_args *args)
 {
 	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
@@ -42,7 +41,7 @@ void pack_geometry_init(struct pack_geometry *geometry,
 			 */
 			continue;
 
-		if (!pack_kept_objects) {
+		if (!args->pack_kept_objects) {
 			/*
 			 * Any pack that has its pack_keep bit set will
 			 * appear in existing->kept_packs below, but
diff --git a/repack.c b/repack.c
index 8a0e4789fa..1982a48165 100644
--- a/repack.c
+++ b/repack.c
@@ -38,6 +38,8 @@ void prepare_pack_objects(struct child_process *cmd,
 		strvec_push(&cmd->args,  "--quiet");
 	if (args->delta_base_offset)
 		strvec_push(&cmd->args,  "--delta-base-offset");
+	if (!args->pack_kept_objects)
+		strvec_push(&cmd->args,  "--honor-pack-keep");
 	strvec_push(&cmd->args, out);
 	cmd->git_cmd = 1;
 	cmd->out = -1;
diff --git a/repack.h b/repack.h
index 9351293233..4a1c4eb606 100644
--- a/repack.h
+++ b/repack.h
@@ -17,10 +17,14 @@ struct pack_objects_args {
 	int name_hash_version;
 	int path_walk;
 	int delta_base_offset;
+	int pack_kept_objects;
 	struct list_objects_filter_options filter_options;
 };
 
-#define PACK_OBJECTS_ARGS_INIT { .delta_base_offset = 1 }
+#define PACK_OBJECTS_ARGS_INIT { \
+	.delta_base_offset = 1, \
+	.pack_kept_objects = -1, \
+}
 
 struct child_process;
 
@@ -104,8 +108,7 @@ struct pack_geometry {
 
 void pack_geometry_init(struct pack_geometry *geometry,
 			struct existing_packs *existing,
-			const struct pack_objects_args *args,
-			int pack_kept_objects);
+			const struct pack_objects_args *args);
 void pack_geometry_split(struct pack_geometry *geometry);
 struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry);
 void pack_geometry_remove_redundant(struct pack_geometry *geometry,
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:10 UTC · re: Taylor Blau · lore

[PATCH 47/49] repack: move `write_filtered_pack()` out of the builtin

In a similar fashion as in previous commits, move the function `write_filtered_pack()` out of the builtin and into its own compilation unit.

This function is now part of the repack.h API, but implemented in its own "repack-filtered.c" unit as it is a separate component from other kinds of repacking operations.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile          |  1 +
 builtin/repack.c  | 46 ------------------------------------------
 meson.build       |  1 +
 repack-filtered.c | 51 +++++++++++++++++++++++++++++++++++++++++++++++
 repack.h          |  4 ++++
 5 files changed, 57 insertions(+), 46 deletions(-)
 create mode 100644 repack-filtered.c
Show changes to 5 files +57 −46

Makefile, builtin/repack.c, meson.build, repack-filtered.c, repack.h

diff --git a/Makefile b/Makefile
index 3360743afb..1e2c09da07 100644
--- a/Makefile
+++ b/Makefile
@@ -1137,6 +1137,7 @@ LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
+LIB_OBJS += repack-filtered.o
 LIB_OBJS += repack-geometry.o
 LIB_OBJS += repack-midx.o
 LIB_OBJS += repack-promisor.o
diff --git a/builtin/repack.c b/builtin/repack.c
index 9d89217b77..a9fc09a24d 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -106,52 +106,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-static int write_filtered_pack(struct write_pack_opts *opts,
-			       struct existing_packs *existing,
-			       struct string_list *names)
-{
-	struct child_process cmd = CHILD_PROCESS_INIT;
-	struct string_list_item *item;
-	FILE *in;
-	int ret;
-	const char *caret;
-	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
-
-	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
-
-	strvec_push(&cmd.args, "--stdin-packs");
-
-	for_each_string_list_item(item, &existing->kept_packs)
-		strvec_pushf(&cmd.args, "--keep-pack=%s", item->string);
-
-	cmd.in = -1;
-
-	ret = start_command(&cmd);
-	if (ret)
-		return ret;
-
-	/*
-	 * Here 'names' contains only the pack(s) that were just
-	 * written, which is exactly the packs we want to keep. Also
-	 * 'existing_kept_packs' already contains the packs in
-	 * 'keep_pack_list'.
-	 */
-	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, names)
-		fprintf(in, "^%s-%s.pack\n", pack_prefix, item->string);
-	for_each_string_list_item(item, &existing->non_kept_packs)
-		fprintf(in, "%s.pack\n", item->string);
-	for_each_string_list_item(item, &existing->cruft_packs)
-		fprintf(in, "%s.pack\n", item->string);
-	caret = opts->po_args->pack_kept_objects ? "" : "^";
-	for_each_string_list_item(item, &existing->kept_packs)
-		fprintf(in, "%s%s.pack\n", caret, item->string);
-	fclose(in);
-
-	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
-				       names);
-}
-
 static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
 				      struct existing_packs *existing)
 {
diff --git a/meson.build b/meson.build
index 42171d1851..4c60e016c3 100644
--- a/meson.build
+++ b/meson.build
@@ -463,6 +463,7 @@ libgit_sources = [
   'reftable/writer.c',
   'remote.c',
   'repack.c',
+  'repack-filtered.c',
   'repack-geometry.c',
   'repack-midx.c',
   'repack-promisor.c',
diff --git a/repack-filtered.c b/repack-filtered.c
new file mode 100644
index 0000000000..96c7b02bb6
--- /dev/null
+++ b/repack-filtered.c
@@ -0,0 +1,51 @@
+#include "git-compat-util.h"
+#include "repack.h"
+#include "repository.h"
+#include "run-command.h"
+#include "string-list.h"
+
+int write_filtered_pack(struct write_pack_opts *opts,
+			struct existing_packs *existing,
+			struct string_list *names)
+{
+	struct child_process cmd = CHILD_PROCESS_INIT;
+	struct string_list_item *item;
+	FILE *in;
+	int ret;
+	const char *caret;
+	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
+
+	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
+
+	strvec_push(&cmd.args, "--stdin-packs");
+
+	for_each_string_list_item(item, &existing->kept_packs)
+		strvec_pushf(&cmd.args, "--keep-pack=%s", item->string);
+
+	cmd.in = -1;
+
+	ret = start_command(&cmd);
+	if (ret)
+		return ret;
+
+	/*
+	 * Here 'names' contains only the pack(s) that were just
+	 * written, which is exactly the packs we want to keep. Also
+	 * 'existing_kept_packs' already contains the packs in
+	 * 'keep_pack_list'.
+	 */
+	in = xfdopen(cmd.in, "w");
+	for_each_string_list_item(item, names)
+		fprintf(in, "^%s-%s.pack\n", pack_prefix, item->string);
+	for_each_string_list_item(item, &existing->non_kept_packs)
+		fprintf(in, "%s.pack\n", item->string);
+	for_each_string_list_item(item, &existing->cruft_packs)
+		fprintf(in, "%s.pack\n", item->string);
+	caret = opts->po_args->pack_kept_objects ? "" : "^";
+	for_each_string_list_item(item, &existing->kept_packs)
+		fprintf(in, "%s%s.pack\n", caret, item->string);
+	fclose(in);
+
+	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
+				       names);
+}
diff --git a/repack.h b/repack.h
index 4a1c4eb606..a7ddbe784b 100644
--- a/repack.h
+++ b/repack.h
@@ -133,4 +133,8 @@ struct repack_write_midx_opts {
 void midx_snapshot_refs(struct repository *repo, struct tempfile *f);
 int write_midx_included_packs(struct repack_write_midx_opts *opts);
 
+int write_filtered_pack(struct write_pack_opts *opts,
+			struct existing_packs *existing,
+			struct string_list *names);
+
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:10 UTC · re: Taylor Blau · lore

[PATCH 48/49] repack: move `write_cruft_pack()` out of the builtin

In an identical fashion as the previous commit, move the function `write_cruft_pack()` into its own compilation unit, and make the function visible through the repack.h API.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile         |  1 +
 builtin/repack.c | 94 ---------------------------------------------
 meson.build      |  1 +
 repack-cruft.c   | 99 ++++++++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  6 +++
 5 files changed, 107 insertions(+), 94 deletions(-)
 create mode 100644 repack-cruft.c
Show changes to 5 files +107 −94

Makefile, builtin/repack.c, meson.build, repack-cruft.c, repack.h

diff --git a/Makefile b/Makefile
index 1e2c09da07..72d7a4686f 100644
--- a/Makefile
+++ b/Makefile
@@ -1137,6 +1137,7 @@ LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
+LIB_OBJS += repack-cruft.o
 LIB_OBJS += repack-filtered.o
 LIB_OBJS += repack-geometry.o
 LIB_OBJS += repack-midx.o
diff --git a/builtin/repack.c b/builtin/repack.c
index a9fc09a24d..9171ca66a7 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -106,100 +106,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
-				      struct existing_packs *existing)
-{
-	struct packfile_store *packs = existing->repo->objects->packfiles;
-	struct packed_git *p;
-	struct strbuf buf = STRBUF_INIT;
-	size_t i;
-
-	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
-		if (!(p->is_cruft && p->pack_local))
-			continue;
-
-		strbuf_reset(&buf);
-		strbuf_addstr(&buf, pack_basename(p));
-		strbuf_strip_suffix(&buf, ".pack");
-
-		if (!string_list_has_string(&existing->cruft_packs, buf.buf))
-			continue;
-
-		if (p->pack_size < combine_cruft_below_size) {
-			fprintf(in, "-%s\n", pack_basename(p));
-		} else {
-			existing_packs_retain_cruft(existing, p);
-			fprintf(in, "%s\n", pack_basename(p));
-		}
-	}
-
-	for (i = 0; i < existing->non_kept_packs.nr; i++)
-		fprintf(in, "-%s.pack\n",
-			existing->non_kept_packs.items[i].string);
-
-	strbuf_release(&buf);
-}
-
-static int write_cruft_pack(struct write_pack_opts *opts,
-			    const char *cruft_expiration,
-			    unsigned long combine_cruft_below_size,
-			    struct string_list *names,
-			    struct existing_packs *existing)
-{
-	struct child_process cmd = CHILD_PROCESS_INIT;
-	struct string_list_item *item;
-	FILE *in;
-	int ret;
-	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
-
-	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
-
-	strvec_push(&cmd.args, "--cruft");
-	if (cruft_expiration)
-		strvec_pushf(&cmd.args, "--cruft-expiration=%s",
-			     cruft_expiration);
-
-	strvec_push(&cmd.args, "--non-empty");
-
-	cmd.in = -1;
-
-	ret = start_command(&cmd);
-	if (ret)
-		return ret;
-
-	/*
-	 * names has a confusing double use: it both provides the list
-	 * of just-written new packs, and accepts the name of the cruft
-	 * pack we are writing.
-	 *
-	 * By the time it is read here, it contains only the pack(s)
-	 * that were just written, which is exactly the set of packs we
-	 * want to consider kept.
-	 *
-	 * If `--expire-to` is given, the double-use served by `names`
-	 * ensures that the pack written to `--expire-to` excludes any
-	 * objects contained in the cruft pack.
-	 */
-	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, names)
-		fprintf(in, "%s-%s.pack\n", pack_prefix, item->string);
-	if (combine_cruft_below_size && !cruft_expiration) {
-		combine_small_cruft_packs(in, combine_cruft_below_size,
-					  existing);
-	} else {
-		for_each_string_list_item(item, &existing->non_kept_packs)
-			fprintf(in, "-%s.pack\n", item->string);
-		for_each_string_list_item(item, &existing->cruft_packs)
-			fprintf(in, "-%s.pack\n", item->string);
-	}
-	for_each_string_list_item(item, &existing->kept_packs)
-		fprintf(in, "%s.pack\n", item->string);
-	fclose(in);
-
-	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
-				       names);
-}
-
 int cmd_repack(int argc,
 	       const char **argv,
 	       const char *prefix,
diff --git a/meson.build b/meson.build
index 4c60e016c3..2993b767dc 100644
--- a/meson.build
+++ b/meson.build
@@ -463,6 +463,7 @@ libgit_sources = [
   'reftable/writer.c',
   'remote.c',
   'repack.c',
+  'repack-cruft.c',
   'repack-filtered.c',
   'repack-geometry.c',
   'repack-midx.c',
diff --git a/repack-cruft.c b/repack-cruft.c
new file mode 100644
index 0000000000..accb98bcdb
--- /dev/null
+++ b/repack-cruft.c
@@ -0,0 +1,99 @@
+#include "git-compat-util.h"
+#include "repack.h"
+#include "packfile.h"
+#include "repository.h"
+#include "run-command.h"
+
+static void combine_small_cruft_packs(FILE *in, off_t combine_cruft_below_size,
+				      struct existing_packs *existing)
+{
+	struct packfile_store *packs = existing->repo->objects->packfiles;
+	struct packed_git *p;
+	struct strbuf buf = STRBUF_INIT;
+	size_t i;
+
+	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
+		if (!(p->is_cruft && p->pack_local))
+			continue;
+
+		strbuf_reset(&buf);
+		strbuf_addstr(&buf, pack_basename(p));
+		strbuf_strip_suffix(&buf, ".pack");
+
+		if (!string_list_has_string(&existing->cruft_packs, buf.buf))
+			continue;
+
+		if (p->pack_size < combine_cruft_below_size) {
+			fprintf(in, "-%s\n", pack_basename(p));
+		} else {
+			existing_packs_retain_cruft(existing, p);
+			fprintf(in, "%s\n", pack_basename(p));
+		}
+	}
+
+	for (i = 0; i < existing->non_kept_packs.nr; i++)
+		fprintf(in, "-%s.pack\n",
+			existing->non_kept_packs.items[i].string);
+
+	strbuf_release(&buf);
+}
+
+int write_cruft_pack(struct write_pack_opts *opts,
+		     const char *cruft_expiration,
+		     unsigned long combine_cruft_below_size,
+		     struct string_list *names,
+		     struct existing_packs *existing)
+{
+	struct child_process cmd = CHILD_PROCESS_INIT;
+	struct string_list_item *item;
+	FILE *in;
+	int ret;
+	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
+
+	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
+
+	strvec_push(&cmd.args, "--cruft");
+	if (cruft_expiration)
+		strvec_pushf(&cmd.args, "--cruft-expiration=%s",
+			     cruft_expiration);
+
+	strvec_push(&cmd.args, "--non-empty");
+
+	cmd.in = -1;
+
+	ret = start_command(&cmd);
+	if (ret)
+		return ret;
+
+	/*
+	 * names has a confusing double use: it both provides the list
+	 * of just-written new packs, and accepts the name of the cruft
+	 * pack we are writing.
+	 *
+	 * By the time it is read here, it contains only the pack(s)
+	 * that were just written, which is exactly the set of packs we
+	 * want to consider kept.
+	 *
+	 * If `--expire-to` is given, the double-use served by `names`
+	 * ensures that the pack written to `--expire-to` excludes any
+	 * objects contained in the cruft pack.
+	 */
+	in = xfdopen(cmd.in, "w");
+	for_each_string_list_item(item, names)
+		fprintf(in, "%s-%s.pack\n", pack_prefix, item->string);
+	if (combine_cruft_below_size && !cruft_expiration) {
+		combine_small_cruft_packs(in, combine_cruft_below_size,
+					  existing);
+	} else {
+		for_each_string_list_item(item, &existing->non_kept_packs)
+			fprintf(in, "-%s.pack\n", item->string);
+		for_each_string_list_item(item, &existing->cruft_packs)
+			fprintf(in, "-%s.pack\n", item->string);
+	}
+	for_each_string_list_item(item, &existing->kept_packs)
+		fprintf(in, "%s.pack\n", item->string);
+	fclose(in);
+
+	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
+				       names);
+}
diff --git a/repack.h b/repack.h
index a7ddbe784b..15886c3488 100644
--- a/repack.h
+++ b/repack.h
@@ -137,4 +137,10 @@ int write_filtered_pack(struct write_pack_opts *opts,
 			struct existing_packs *existing,
 			struct string_list *names);
 
+int write_cruft_pack(struct write_pack_opts *opts,
+		     const char *cruft_expiration,
+		     unsigned long combine_cruft_below_size,
+		     struct string_list *names,
+		     struct existing_packs *existing);
+
 #endif /* REPACK_H */
-- 
2.51.0.243.g16eca91f2c0
Taylor Blau· Sep 28, 2025, 22:10 UTC · re: Taylor Blau · lore

[PATCH 49/49] builtin/repack.c: clean up unused `#include`s

Over the past several dozen commits, we have moved a large amount of functionality out of the repack builtin and into other files like repack.c, repack-cruft.c, repack-filtered.c, repack-midx.c, and repack-promisor.c.

These files specify the minimal set of `#include`s that they need to compile successfully, but we did not change the set of `#include`s in the repack builtin itself.

Now that the code movement is complete, let's clean up that set of `#include`s and trim down the builtin to include the minimal amount of external headers necessary to compile.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 9 ---------
 1 file changed, 9 deletions(-)
Show changes to builtin/repack.c +0 −9
diff --git a/builtin/repack.c b/builtin/repack.c
index 9171ca66a7..ad60c4290d 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -4,26 +4,17 @@
 #include "builtin.h"
 #include "config.h"
 #include "environment.h"
-#include "gettext.h"
-#include "hex.h"
 #include "parse-options.h"
 #include "path.h"
 #include "run-command.h"
 #include "server-info.h"
-#include "strbuf.h"
 #include "string-list.h"
-#include "strvec.h"
 #include "midx.h"
 #include "packfile.h"
 #include "prune-packed.h"
-#include "odb.h"
 #include "promisor-remote.h"
 #include "repack.h"
 #include "shallow.h"
-#include "pack.h"
-#include "pack-bitmap.h"
-#include "refs.h"
-#include "list-objects-filter-options.h"
 
 #define ALL_INTO_ONE 1
 #define LOOSEN_UNREACHABLE 2
-- 
2.51.0.243.g16eca91f2c0
Junio C Hamano· Sep 28, 2025, 22:58 UTC · re: Taylor Blau · lore

Re: [PATCH 00/49] repack: prepare for incremental MIDX-based repacking

Taylor Blau <me@ttaylorr.com> writes:
Show 7 quoted lines
>  * This series depends on ps/packfile-store at dd52a29b78 (packfile:
>    refactor `get_packed_git_mru()` to work on packfile store,
>    2025-09-23).
>
>  * This series is split out of my local copy of
>    tb/incremental-midx-part-3. I suggest queueing it as
>    tb/incremental-midx-part-3.1 ;-).

Both of these comments are very much appreciated. Do we say something to encourage these in our developer-facing documentation?

The points to stress are:
 - If you are not building directly on top of the recent tip of
   'master', mention what topics that are not yet in 'master' your
   patches depend on.
 - If your series is a part of a larger effort, briefly describe
   what your overall direction and grand vision is, and state where
   in that grand picture the current series fits.
 - If you have a short one-line summary for the topic to be used as
   a topic branch name in mind, do suggest it.
Thanks.
Taylor Blau· Oct 7, 2025, 20:44 UTC · re: Junio C Hamano · lore

Re: [PATCH 00/49] repack: prepare for incremental MIDX-based repacking

On Sun, Sep 28, 2025 at 03:58:00PM -0700, Junio C Hamano wrote:
Show 12 quoted lines
> Taylor Blau <me@ttaylorr.com> writes:
>
> >  * This series depends on ps/packfile-store at dd52a29b78 (packfile:
> >    refactor `get_packed_git_mru()` to work on packfile store,
> >    2025-09-23).
> >
> >  * This series is split out of my local copy of
> >    tb/incremental-midx-part-3. I suggest queueing it as
> >    tb/incremental-midx-part-3.1 ;-).
>
> Both of these comments are very much appreciated.  Do we say
> something to encourage these in our developer-facing documentation?
Thanks.
Show 5 quoted lines
> The points to stress are:
>
>  - If you are not building directly on top of the recent tip of
>    'master', mention what topics that are not yet in 'master' your
>    patches depend on.

I think that this one is covered in general by 0a02ca2383 (SubmittingPatches: simplify guidance for choosing a starting point, 2023-07-14). The language added there is a little more generic, since it suggests mentioning how you constructed a base branch when a new topic depends on >1 other topics.

I would imagine it's clear to readers there that "if I am describing how to build a base branch which merges in multiple topics, I should likewise list the topic branch by name if I am building on top of a single one already in the maintainer's tree."

>  - If your series is a part of a larger effort, briefly describe
>    what your overall direction and grand vision is, and state where
>    in that grand picture the current series fits.
Not covered.
>  - If you have a short one-line summary for the topic to be used as
>    a topic branch name in mind, do suggest it.

Sort of covered by your d255105c99 (SubmittingPatches: release-notes entry experiment, 2024-03-25), but that talks more about the summary for RelNotes purposes, not the topic branch name.

I'll send out a short series to add some more guidelines to cover the above in our Documentation/SubmittingPatches.

> Thanks.

Thanks, Taylor

Jeff King· Oct 10, 2025, 06:29 UTC · re: Taylor Blau · lore

Re: [PATCH 00/49] repack: prepare for incremental MIDX-based repacking

On Sun, Sep 28, 2025 at 06:07:10PM -0400, Taylor Blau wrote:
Show 9 quoted lines
> This series moves the vast majority of the builtin's functionality into
> several new compilation units, all grouped under a new repack.h API:
> 
>  * repack.c
>  * repack-cruft.c
>  * repack-filtered.c
>  * repack-geometry.c
>  * repack-midx.c
>  * repack-promisor.c

I've read up through patch 37. My eyes were starting to glaze over at that point, and it looked like the patches were starting to get more interesting, so I've saved the rest for another session.

From what I read, the changes all look reasonable. I brought up a few small naming and interface questions, but I don't think there's anything incorrect here. And there may be diminishing returns in polishing an API which is realistically only going to be called from one place.

The split you have seems reasonable. It's not quite as clean as one might hope (e.g., repack's existing_packs gets passed around to some of the other subsystems). But again, diminishing returns on trying to untangle it further.

-Peff
Taylor Blau· Oct 10, 2025, 23:15 UTC · re: Jeff King · lore

Re: [PATCH 00/49] repack: prepare for incremental MIDX-based repacking

On Fri, Oct 10, 2025 at 02:29:13AM -0400, Jeff King wrote:
Show 15 quoted lines
> On Sun, Sep 28, 2025 at 06:07:10PM -0400, Taylor Blau wrote:
>
> > This series moves the vast majority of the builtin's functionality into
> > several new compilation units, all grouped under a new repack.h API:
> >
> >  * repack.c
> >  * repack-cruft.c
> >  * repack-filtered.c
> >  * repack-geometry.c
> >  * repack-midx.c
> >  * repack-promisor.c
>
> I've read up through patch 37. My eyes were starting to glaze over at
> that point, and it looked like the patches were starting to get more
> interesting, so I've saved the rest for another session.
Thanks for the review thus far!
Show 9 quoted lines
> From what I read, the changes all look reasonable. I brought up a few
> small naming and interface questions, but I don't think there's anything
> incorrect here. And there may be diminishing returns in polishing an API
> which is realistically only going to be called from one place.
>
> The split you have seems reasonable. It's not quite as clean as one
> might hope (e.g., repack's existing_packs gets passed around to some of
> the other subsystems). But again, diminishing returns on trying to
> untangle it further.

Yeah, I think we could probably go on for quite a bit longer polishing the naming, interfaces, etc. I tried to strike a balance in this series to change as little as possible while still ending up with a halfway-decent looking repack.h ;-).

As you noted, the repack code is extremely fragile/brittle. That is a big part of why I both (a) took such an incremental approach as in this series and (b) tried to avoid making too many changes as I moved code around.

I think that (b) especially leaves us with an opportunity to clean these interfaces up after this series has landed, and I intend on doing so. My hope is that this series gives us a solid foundation to make those changes on, as opposed to further hacking up builtin/repack.c ;-).

Thanks, Taylor

Jeff King· Oct 15, 2025, 11:06 UTC · re: Taylor Blau · lore

Re: [PATCH 00/49] repack: prepare for incremental MIDX-based repacking

On Fri, Oct 10, 2025 at 07:15:54PM -0400, Taylor Blau wrote:
Show 5 quoted lines
> > I've read up through patch 37. My eyes were starting to glaze over at
> > that point, and it looked like the patches were starting to get more
> > interesting, so I've saved the rest for another session.
> 
> Thanks for the review thus far!

OK, I've finished reading through. The remainder was actually less complicated than I thought it might be.

I left a few comments, but there's nothing really important. It sounds like you're doing a re-roll for minor fixes anyway, so you can take or leave my ponderings as you see fit.

Show 14 quoted lines
> Yeah, I think we could probably go on for quite a bit longer polishing
> the naming, interfaces, etc. I tried to strike a balance in this series
> to change as little as possible while still ending up with a
> halfway-decent looking repack.h ;-).
> 
> As you noted, the repack code is extremely fragile/brittle. That is a
> big part of why I both (a) took such an incremental approach as in this
> series and (b) tried to avoid making too many changes as I moved code
> around.
> 
> I think that (b) especially leaves us with an opportunity to clean these
> interfaces up after this series has landed, and I intend on doing so. My
> hope is that this series gives us a solid foundation to make those
> changes on, as opposed to further hacking up builtin/repack.c ;-).

OK, that all sounds good to me. I think even if you never circle back to clean up further, the end result here is a net improvement.

-Peff
Taylor Blau· Oct 15, 2025, 22:26 UTC · re: Taylor Blau · lore

[PATCH v2 00/49] repack: prepare for incremental MIDX-based repacking

Note to the maintainer:
 * This series has been rebased on to 'master' (which is 4b71b29477
   (The seventeenth batch, 2025-10-10) at the time of writing) to
   avoid semantic conflicts now that ps/packfile-store has been
   merged.

This is a relatively small reroll of my series to clean up the repack builtin by introducing a repack.h API, which is the first of ~three series that will implement incremental MIDX/bitmap-based repacking.

A range-diff is included below for convenience, but the changes since v1 are generally limited to the following:

 * Wording tweaks and a couple of minor typo fixes.
 * Dropping explicit casts out of 'void *'.
 * Clarification in commit "builtin/repack.c: introduce `struct
   write_pack_opts`" that additional cleanup follows in the upcoming
   patches.
 * Marking parts of the new API as const where possible.
 * Using 'bool' as the return type and simplifying the implementation
   of `write_pack_opts_is_local()`.
 * Avoid shadowing "struct write_pack_opts opts" in `cmd_repack()`.

Outside of that, the series is unchanged, and I am hopeful that this round looks good to reviewers so that we can move on to the more interesting parts of incremental MIDX/bitmap repacking ;-).

Thanks in advance for your (re-)review!
Taylor Blau (49):
  builtin/repack.c: avoid "the_repository" in `cmd_repack()`
  builtin/repack.c: avoid "the_repository" in existing packs API
  builtin/repack.c: avoid "the_repository" when taking a ref snapshot
  builtin/repack.c: avoid "the_repository" when removing packs
  builtin/repack.c: avoid "the_repository" when repacking promisor
    objects
  builtin/repack.c: avoid "the_hash_algo" when deleting packs
  builtin/repack.c: avoid "the_hash_algo" in `write_oid()`
  builtin/repack: avoid "the_hash_algo" in `repack_promisor_objects()`
  builtin/repack.c: avoid "the_hash_algo" in `finish_pack_objects_cmd()`
  builtin/repack.c: avoid using `hash_to_hex()` in pack geometry
  repack: introduce new compilation unit
  builtin/repack.c: pass both pack_objects args to repack_config
  repack: move 'delta_base_offset' to 'struct pack_objects_args'
  repack: remove 'prepare_pack_objects' from the builtin
  builtin/repack.c: rename many 'struct existing_packs' functions
  repack: remove 'remove_redundant_pack' from the builtin
  builtin/repack.c: pass "packdir" when removing packs
  builtin/repack.c: avoid unnecessary numeric casts in existing_packs
  repack: remove 'existing_packs' API from the builtin
  builtin/repack.c: rename "struct generated_pack_data"
  builtin/repack.c: factor out "generated_pack_install"
  builtin/repack.c: pass "packtmp" to `generated_pack_populate()`
  builtin/repack.c: provide pack locations to `generated_pack_install()`
  repack: remove 'generated_pack' API from the builtin
  builtin/repack.c: pass "packtmp" to `repack_promisor_objects()`
  builtin/repack.c: remove "repack_promisor_objects()" from the builtin
  builtin/repack.c: rename various pack_geometry functions
  builtin/repack.c: pass 'pack_kept_objects' to `pack_geometry_init()`
  builtin/repack.c: pass 'packdir' to `pack_geometry_remove_redundant()`
  repack: remove pack_geometry API from the builtin
  builtin/repack.c: remove ref snapshotting from builtin
  builtin/repack.c: extract opts struct for
    'write_midx_included_packs()'
  builtin/repack.c: use a string_list for 'midx_pack_names'
  repack: keep track of MIDX pack names using existing_packs
  builtin/repack.c: reorder `remove_redundant_bitmaps()`
  builtin/repack.c: inline `remove_redundant_bitmaps()`
  builtin/repack.c: pass `repack_write_midx_opts` to
    `midx_included_packs`
  builtin/repack.c: inline packs within `write_midx_included_packs()`
  repack: 'write_midx_included_packs' API from the builtin
  builtin/repack.c: introduce `struct write_pack_opts`
  builtin/repack.c: use `write_pack_opts` within `write_cruft_pack()`
  repack: move `find_pack_prefix()` out of the builtin
  repack: extract `write_pack_opts_is_local()`
  builtin/repack.c: pass `write_pack_opts` to
    `finish_pack_objects_cmd()`
  repack: move `finish_pack_objects_cmd()` out of the builtin
  repack: move `pack_kept_objects` to `struct pack_objects_args`
  repack: move `write_filtered_pack()` out of the builtin
  repack: move `write_cruft_pack()` out of the builtin
  builtin/repack.c: clean up unused `#include`s
 Makefile          |    6 +
 builtin/repack.c  | 1356 +++------------------------------------------
 meson.build       |    6 +
 repack-cruft.c    |   99 ++++
 repack-filtered.c |   51 ++
 repack-geometry.c |  233 ++++++++
 repack-midx.c     |  372 +++++++++++++
 repack-promisor.c |  102 ++++
 repack.c          |  360 ++++++++++++
 repack.h          |  146 +++++
 10 files changed, 1467 insertions(+), 1264 deletions(-)
 create mode 100644 repack-cruft.c
 create mode 100644 repack-filtered.c
 create mode 100644 repack-geometry.c
 create mode 100644 repack-midx.c
 create mode 100644 repack-promisor.c
 create mode 100644 repack.c
 create mode 100644 repack.h
Range-diff against v1:
 1:  aa03aea8ed =  1:  fc2b686c49 builtin/repack.c: avoid "the_repository" in `cmd_repack()`
 2:  664a67c936 =  2:  14a65d7c6d builtin/repack.c: avoid "the_repository" in existing packs API
 3:  866e43b0cd =  3:  ff72cd499b builtin/repack.c: avoid "the_repository" when taking a ref snapshot
 4:  f404dfa34f =  4:  75f23ba211 builtin/repack.c: avoid "the_repository" when removing packs
 5:  6190eeb066 =  5:  0914ca113f builtin/repack.c: avoid "the_repository" when repacking promisor objects
 6:  2f97882ace =  6:  238f3d87a0 builtin/repack.c: avoid "the_hash_algo" when deleting packs
 7:  37a7dea848 =  7:  ca3c8a782a builtin/repack.c: avoid "the_hash_algo" in `write_oid()`
 8:  f5a87f2a65 =  8:  34b503618a builtin/repack: avoid "the_hash_algo" in `repack_promisor_objects()`
 9:  f16dfbf0c7 =  9:  cd43a60620 builtin/repack.c: avoid "the_hash_algo" in `finish_pack_objects_cmd()`
10:  de4886466c = 10:  c6787c865d builtin/repack.c: avoid using `hash_to_hex()` in pack geometry
11:  fe3328eb3d = 11:  87f2903e75 repack: introduce new compilation unit
12:  75386eed60 = 12:  a812bd3dcb builtin/repack.c: pass both pack_objects args to repack_config
13:  06347b6ab2 = 13:  f2d1fa51e3 repack: move 'delta_base_offset' to 'struct pack_objects_args'
14:  2d9022385d = 14:  0c32c3c9b6 repack: remove 'prepare_pack_objects' from the builtin
15:  02e26f78ac = 15:  a9767dc42d builtin/repack.c: rename many 'struct existing_packs' functions
16:  d0a40e8e57 = 16:  4edbcb5893 repack: remove 'remove_redundant_pack' from the builtin
17:  df75c3130b = 17:  4e2a152ebb builtin/repack.c: pass "packdir" when removing packs
18:  f413f958eb = 18:  46aa154b76 builtin/repack.c: avoid unnecessary numeric casts in existing_packs
19:  ae267fcc82 ! 19:  0c6dea2800 repack: remove 'existing_packs' API from the builtin
    @@ Commit message
         end of a repack operation.
     
         Now that the prerequisite refactoring is complete, this API is isolated
    -    enough that it can be moved out to repack.ch and removed from the
    +    enough that it can be moved out to repack.[ch] and removed from the
         builtin entirely.
     
         As a result, some of its functions become static within repack.c,
20:  ee3e1233f5 = 20:  53d5b1053f builtin/repack.c: rename "struct generated_pack_data"
21:  96d27261e7 ! 21:  196b92d052 builtin/repack.c: factor our "generated_pack_install"
    @@ Metadata
     Author: Taylor Blau <me@ttaylorr.com>
     
      ## Commit message ##
    -    builtin/repack.c: factor our "generated_pack_install"
    +    builtin/repack.c: factor out "generated_pack_install"
     
         Once all new packs are known to exist, 'repack' installs their contents
         from their temporary location into their permanent one. This is a
    @@ builtin/repack.c: int cmd_repack(int argc,
     -		}
     -	}
     +	for_each_string_list_item(item, &names)
    -+		generated_pack_install((struct generated_pack *)item->util,
    -+				       item->string);
    ++		generated_pack_install(item->util, item->string);
      	/* End of pack replacement. */
      
      	if (delete_redundant && pack_everything & ALL_INTO_ONE)
22:  ab0dc66683 = 22:  acd9a83ed0 builtin/repack.c: pass "packtmp" to `generated_pack_populate()`
23:  e9d525d83a ! 23:  2b81a26a84 builtin/repack.c: provide pack locations to `generated_pack_install()`
    @@ builtin/repack.c: static int generated_pack_has_ext(const struct generated_pack
      	int ext;
      	for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
     @@ builtin/repack.c: int cmd_repack(int argc,
    + 	 * Ok we have prepared all new packfiles.
      	 */
      	for_each_string_list_item(item, &names)
    - 		generated_pack_install((struct generated_pack *)item->util,
    --				       item->string);
    -+				       item->string, packdir, packtmp);
    +-		generated_pack_install(item->util, item->string);
    ++		generated_pack_install(item->util, item->string, packdir,
    ++				       packtmp);
      	/* End of pack replacement. */
      
      	if (delete_redundant && pack_everything & ALL_INTO_ONE)
24:  9979b77979 = 24:  69003f4561 repack: remove 'generated_pack' API from the builtin
25:  67effcb4ea = 25:  ca030322cf builtin/repack.c: pass "packtmp" to `repack_promisor_objects()`
26:  3ba6e86f61 ! 26:  1050b856a5 builtin/repack.c: remove "repack_promisor_objects()" from the builtin
    @@ repack-promisor.c (new)
     @@
     +#include "git-compat-util.h"
     +#include "repack.h"
    -+#include "run-command.h"
     +#include "hex.h"
    -+#include "repository.h"
    ++#include "pack.h"
     +#include "packfile.h"
     +#include "path.h"
    -+#include "pack.h"
    ++#include "repository.h"
    ++#include "run-command.h"
     +
     +struct write_oid_context {
     +	struct child_process *cmd;
27:  dd67366476 ! 27:  0b2fa0cf62 builtin/repack.c: rename various pack_geometry functions
    @@ Commit message
         builtin/repack.c: rename various pack_geometry functions
     
         Rename functions which work with 'struct pack_geometry' to begin with
    -    "pack_geometry_".
    +    "pack_geometry_". While we're at it, change `free_pack_geometry()` to
    +    instead be named `pack_geometry_release()` to match our conventions, and
    +    make clear that that function frees the contents of the struct, not the
    +    memory allocated to hold the struct itself.
     
         Signed-off-by: Taylor Blau <me@ttaylorr.com>
     
28:  8408fd8474 = 28:  a3d99c4852 builtin/repack.c: pass 'pack_kept_objects' to `pack_geometry_init()`
29:  76be98b386 = 29:  0a2a72ea22 builtin/repack.c: pass 'packdir' to `pack_geometry_remove_redundant()`
30:  54a2feaea9 ! 30:  9172e9a515 repack: remove pack_geometry API from the builtin
    @@ repack-geometry.c (new)
     +
     +#include "git-compat-util.h"
     +#include "repack.h"
    ++#include "repository.h"
     +#include "hex.h"
     +#include "packfile.h"
     +
31:  06ca4ab5b7 = 31:  5b15d83e65 builtin/repack.c: remove ref snapshotting from builtin
32:  3b7e204a40 = 32:  077f06c935 builtin/repack.c: extract opts struct for 'write_midx_included_packs()'
33:  8844529759 = 33:  40ce496de2 builtin/repack.c: use a string_list for 'midx_pack_names'
34:  ccf2c6a517 = 34:  ec3a8a7dd2 repack: keep track of MIDX pack names using existing_packs
35:  85d229fff5 = 35:  8042860b26 builtin/repack.c: reorder `remove_redundant_bitmaps()`
36:  2bf732009a = 36:  af06f60e63 builtin/repack.c: inline `remove_redundant_bitmaps()`
37:  60659a7549 = 37:  668f4edc9e builtin/repack.c: pass `repack_write_midx_opts` to `midx_included_packs`
38:  c2c7ca9f9e ! 38:  5bd43e96f2 builtin/repack.c: inline packs within `write_midx_included_packs()`
    @@ Commit message
                 write_midx_included_packs(&included_packs, ...);
             }
     
    -    in the future, let's have each function which writes a MIDX be
    +    in the future, let's have each function that writes a MIDX be
         responsible for itself computing the list of included packs. Inline the
         declaration and initialization of `included_packs` into the
         `write_midx_included_packs()` function itself, and repeat that pattern
39:  c28891443b = 39:  794122266d repack: 'write_midx_included_packs' API from the builtin
40:  1091096186 ! 40:  0cb6e78856 builtin/repack.c: introduce `struct write_pack_opts`
    @@ Commit message
         of the same type, and provide a unified interface for the two functions
         mentioned earlier.
     
    +    (Note that "pack_prefix" can be derived on the fly as a function of
    +    "packdir" and "packtmp", making it unnecessary to store "pack_prefix"
    +    explicitly. This commit ignores that potential cleanup in the name of
    +    doing as few things as possible, but a later commit will make that
    +    change.)
    +
         Signed-off-by: Taylor Blau <me@ttaylorr.com>
     
      ## builtin/repack.c ##
    @@ builtin/repack.c: static int finish_pack_objects_cmd(const struct git_hash_algo
     -static int write_filtered_pack(const struct pack_objects_args *args,
     -			       const char *destination,
     -			       const char *pack_prefix,
    -+static int write_filtered_pack(struct write_pack_opts *opts,
    ++static int write_filtered_pack(const struct write_pack_opts *opts,
      			       struct existing_packs *existing,
      			       struct string_list *names)
      {
41:  7826afc173 ! 41:  1a9b578fdd builtin/repack.c: use `write_pack_opts` within `write_cruft_pack()`
    @@ builtin/repack.c: static void combine_small_cruft_packs(FILE *in, size_t combine
     -static int write_cruft_pack(const struct pack_objects_args *args,
     -			    const char *destination,
     -			    const char *pack_prefix,
    -+static int write_cruft_pack(struct write_pack_opts *opts,
    ++static int write_cruft_pack(const struct write_pack_opts *opts,
      			    const char *cruft_expiration,
      			    unsigned long combine_cruft_below_size,
      			    struct string_list *names,
42:  c50ffce078 ! 42:  43116b8942 repack: move `find_pack_prefix()` out of the builtin
    @@ Commit message
         Signed-off-by: Taylor Blau <me@ttaylorr.com>
     
      ## builtin/repack.c ##
    -@@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_filtered_pack(const struct write_pack_opts *opts,
      	const char *caret;
      	const char *scratch;
      	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
    @@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
      
      	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
      
    -@@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_filtered_pack(const struct write_pack_opts *opts,
      	 */
      	in = xfdopen(cmd.in, "w");
      	for_each_string_list_item(item, names)
    @@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
      	for_each_string_list_item(item, &existing->non_kept_packs)
      		fprintf(in, "%s.pack\n", item->string);
      	for_each_string_list_item(item, &existing->cruft_packs)
    -@@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_cruft_pack(const struct write_pack_opts *opts,
      	int ret;
      	const char *scratch;
      	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
    @@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
      
      	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
      
    -@@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_cruft_pack(const struct write_pack_opts *opts,
      	 */
      	in = xfdopen(cmd.in, "w");
      	for_each_string_list_item(item, names)
    @@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
      	if (combine_cruft_below_size && !cruft_expiration) {
      		combine_small_cruft_packs(in, combine_cruft_below_size,
      					  existing);
    -@@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_cruft_pack(const struct write_pack_opts *opts,
      				       local);
      }
      
    @@ repack.c: void repack_remove_redundant_pack(struct repository *repo, const char
      	strbuf_release(&buf);
      }
      
    -+const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts)
    ++const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts)
     +{
     +	const char *pack_prefix;
     +	if (!skip_prefix(opts->packtmp, opts->packdir, &pack_prefix))
    @@ repack.h: void repack_remove_redundant_pack(struct repository *repo, const char
      	const char *packtmp;
      };
      
    -+const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts);
    ++const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts);
     +
      struct repository;
      struct packed_git;
43:  1fbe3d0d7f ! 43:  5e3b615993 repack: extract `write_pack_opts_is_local()`
    @@ Commit message
     
         Extract that function and make it visible within the repack.h API, and
         use it within both `write_cruft_pack()` and `write_filtered_pack()`.
    +    While we're at it, match our modern conventions by returning a "bool"
    +    instead of "int", and use `starts_with()` instead of `skip_prefix()` to
    +    avoid storing the dummy "scratch" variable.
     
         The remaining duplication (that is, that both `write_cruft_pack()` and
         `write_filtered_pack()` still both call `write_pack_opts_is_local()`)
    @@ Commit message
         Signed-off-by: Taylor Blau <me@ttaylorr.com>
     
      ## builtin/repack.c ##
    -@@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_filtered_pack(const struct write_pack_opts *opts,
      	FILE *in;
      	int ret;
      	const char *caret;
     -	const char *scratch;
     -	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
    -+	int local = write_pack_opts_is_local(opts);
    ++	bool local = write_pack_opts_is_local(opts);
      	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
      
      	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
    -@@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_cruft_pack(const struct write_pack_opts *opts,
      	struct string_list_item *item;
      	FILE *in;
      	int ret;
     -	const char *scratch;
     -	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
    -+	int local = write_pack_opts_is_local(opts);
    ++	bool local = write_pack_opts_is_local(opts);
      	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
      
      	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
     
      ## repack.c ##
    -@@ repack.c: const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts)
    +@@ repack.c: const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts)
      	return pack_prefix;
      }
      
    -+int write_pack_opts_is_local(struct write_pack_opts *opts)
    ++bool write_pack_opts_is_local(const struct write_pack_opts *opts)
     +{
    -+	const char *scratch;
    -+	return skip_prefix(opts->destination, opts->packdir, &scratch);
    ++	return starts_with(opts->destination, opts->packdir);
     +}
     +
      #define DELETE_PACK 1
    @@ repack.h
     @@ repack.h: struct write_pack_opts {
      };
      
    - const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts);
    -+int write_pack_opts_is_local(struct write_pack_opts *opts);
    + const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts);
    ++bool write_pack_opts_is_local(const struct write_pack_opts *opts);
      
      struct repository;
      struct packed_git;
44:  11f78993f1 ! 44:  fcbe44d8bb builtin/repack.c: pass `write_pack_opts` to `finish_pack_objects_cmd()`
    @@ builtin/repack.c: static int repack_config(const char *var, const char *value,
      }
      
      static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
    -+				   struct write_pack_opts *opts,
    ++				   const struct write_pack_opts *opts,
      				   struct child_process *cmd,
     -				   struct string_list *names,
     -				   int local)
     +				   struct string_list *names)
      {
      	FILE *out;
    -+	int local = write_pack_opts_is_local(opts);
    ++	bool local = write_pack_opts_is_local(opts);
      	struct strbuf line = STRBUF_INIT;
      
      	out = xfdopen(cmd->out, "r");
    @@ builtin/repack.c: static int finish_pack_objects_cmd(const struct git_hash_algo
      		}
      	}
      	fclose(out);
    -@@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_filtered_pack(const struct write_pack_opts *opts,
      	FILE *in;
      	int ret;
      	const char *caret;
    --	int local = write_pack_opts_is_local(opts);
    +-	bool local = write_pack_opts_is_local(opts);
      	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
      
      	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
    -@@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_filtered_pack(const struct write_pack_opts *opts,
      		fprintf(in, "%s%s.pack\n", caret, item->string);
      	fclose(in);
      
    @@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
      }
      
      static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
    -@@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_cruft_pack(const struct write_pack_opts *opts,
      	struct string_list_item *item;
      	FILE *in;
      	int ret;
    --	int local = write_pack_opts_is_local(opts);
    +-	bool local = write_pack_opts_is_local(opts);
      	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
      
      	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
    -@@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_cruft_pack(const struct write_pack_opts *opts,
      		fprintf(in, "%s.pack\n", item->string);
      	fclose(in);
      
    @@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
      }
      
      int cmd_repack(int argc,
    -@@ builtin/repack.c: int cmd_repack(int argc,
    - 	struct existing_packs existing = EXISTING_PACKS_INIT;
    - 	struct pack_geometry geometry = { 0 };
    - 	struct tempfile *refs_snapshot = NULL;
    -+	struct write_pack_opts opts = { 0 };
    - 	int i, ret;
    - 	int show_progress;
    - 
     @@ builtin/repack.c: int cmd_repack(int argc,
      		fclose(in);
      	}
      
     -	ret = finish_pack_objects_cmd(repo->hash_algo, &cmd, &names, 1);
    -+	opts.packdir = packdir;
    -+	opts.destination = packdir;
    -+	opts.packtmp = packtmp;
    -+	ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd, &names);
    - 	if (ret)
    - 		goto cleanup;
    +-	if (ret)
    +-		goto cleanup;
    ++	{
    ++		struct write_pack_opts opts = {
    ++			.packdir = packdir,
    ++			.destination = packdir,
    ++			.packtmp = packtmp,
    ++		};
    ++		ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd,
    ++					      &names);
    ++		if (ret)
    ++			goto cleanup;
    ++	}
      
    + 	if (!names.nr) {
    + 		if (!po_args.quiet)
45:  a38d14e411 ! 45:  244260c6ab repack: move `finish_pack_objects_cmd()` out of the builtin
    @@ builtin/repack.c: static int repack_config(const char *var, const char *value,
      }
      
     -static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
    --				   struct write_pack_opts *opts,
    +-				   const struct write_pack_opts *opts,
     -				   struct child_process *cmd,
     -				   struct string_list *names)
     -{
     -	FILE *out;
    --	int local = write_pack_opts_is_local(opts);
    +-	bool local = write_pack_opts_is_local(opts);
     -	struct strbuf line = STRBUF_INIT;
     -
     -	out = xfdopen(cmd->out, "r");
    @@ builtin/repack.c: static int repack_config(const char *var, const char *value,
     -	return finish_command(cmd);
     -}
     -
    - static int write_filtered_pack(struct write_pack_opts *opts,
    + static int write_filtered_pack(const struct write_pack_opts *opts,
      			       struct existing_packs *existing,
      			       struct string_list *names)
     
      ## repack.c ##
    -@@ repack.c: int write_pack_opts_is_local(struct write_pack_opts *opts)
    - 	return skip_prefix(opts->destination, opts->packdir, &scratch);
    +@@ repack.c: bool write_pack_opts_is_local(const struct write_pack_opts *opts)
    + 	return starts_with(opts->destination, opts->packdir);
      }
      
     +int finish_pack_objects_cmd(const struct git_hash_algo *algop,
    -+			    struct write_pack_opts *opts,
    ++			    const struct write_pack_opts *opts,
     +			    struct child_process *cmd,
     +			    struct string_list *names)
     +{
     +	FILE *out;
    -+	int local = write_pack_opts_is_local(opts);
    ++	bool local = write_pack_opts_is_local(opts);
     +	struct strbuf line = STRBUF_INIT;
     +
     +	out = xfdopen(cmd->out, "r");
    @@ repack.c: int write_pack_opts_is_local(struct write_pack_opts *opts)
     
      ## repack.h ##
     @@ repack.h: struct write_pack_opts {
    - const char *write_pack_opts_pack_prefix(struct write_pack_opts *opts);
    - int write_pack_opts_is_local(struct write_pack_opts *opts);
    + const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts);
    + bool write_pack_opts_is_local(const struct write_pack_opts *opts);
      
     +int finish_pack_objects_cmd(const struct git_hash_algo *algop,
    -+			    struct write_pack_opts *opts,
    ++			    const struct write_pack_opts *opts,
     +			    struct child_process *cmd,
     +			    struct string_list *names);
     +
46:  2d9a879bb0 ! 46:  dbc317af27 repack: move `pack_kept_objects` to `struct pack_objects_args`
    @@ builtin/repack.c: static int repack_config(const char *var, const char *value,
      		return 0;
      	}
      	if (!strcmp(var, "repack.writebitmaps") ||
    -@@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_filtered_pack(const struct write_pack_opts *opts,
      
      	strvec_push(&cmd.args, "--stdin-packs");
      
    @@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
      	for_each_string_list_item(item, &existing->kept_packs)
      		strvec_pushf(&cmd.args, "--keep-pack=%s", item->string);
      
    -@@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_filtered_pack(const struct write_pack_opts *opts,
      		fprintf(in, "%s.pack\n", item->string);
      	for_each_string_list_item(item, &existing->cruft_packs)
      		fprintf(in, "%s.pack\n", item->string);
    @@ builtin/repack.c: static int write_filtered_pack(struct write_pack_opts *opts,
      	for_each_string_list_item(item, &existing->kept_packs)
      		fprintf(in, "%s%s.pack\n", caret, item->string);
      	fclose(in);
    -@@ builtin/repack.c: static int write_cruft_pack(struct write_pack_opts *opts,
    +@@ builtin/repack.c: static int write_cruft_pack(const struct write_pack_opts *opts,
      		strvec_pushf(&cmd.args, "--cruft-expiration=%s",
      			     cruft_expiration);
      
47:  d0df139717 ! 47:  7523345056 repack: move `write_filtered_pack()` out of the builtin
    @@ builtin/repack.c: static int repack_config(const char *var, const char *value,
      	return git_default_config(var, value, ctx, cb);
      }
      
    --static int write_filtered_pack(struct write_pack_opts *opts,
    +-static int write_filtered_pack(const struct write_pack_opts *opts,
     -			       struct existing_packs *existing,
     -			       struct string_list *names)
     -{
    @@ repack-filtered.c (new)
     +#include "run-command.h"
     +#include "string-list.h"
     +
    -+int write_filtered_pack(struct write_pack_opts *opts,
    ++int write_filtered_pack(const struct write_pack_opts *opts,
     +			struct existing_packs *existing,
     +			struct string_list *names)
     +{
    @@ repack.h: struct repack_write_midx_opts {
      void midx_snapshot_refs(struct repository *repo, struct tempfile *f);
      int write_midx_included_packs(struct repack_write_midx_opts *opts);
      
    -+int write_filtered_pack(struct write_pack_opts *opts,
    ++int write_filtered_pack(const struct write_pack_opts *opts,
     +			struct existing_packs *existing,
     +			struct string_list *names);
     +
48:  4c6ec504e3 ! 48:  567ccd3d3e repack: move `write_cruft_pack()` out of the builtin
    @@ builtin/repack.c: static int repack_config(const char *var, const char *value,
     -	strbuf_release(&buf);
     -}
     -
    --static int write_cruft_pack(struct write_pack_opts *opts,
    +-static int write_cruft_pack(const struct write_pack_opts *opts,
     -			    const char *cruft_expiration,
     -			    unsigned long combine_cruft_below_size,
     -			    struct string_list *names,
    @@ repack-cruft.c (new)
     +	strbuf_release(&buf);
     +}
     +
    -+int write_cruft_pack(struct write_pack_opts *opts,
    ++int write_cruft_pack(const struct write_pack_opts *opts,
     +		     const char *cruft_expiration,
     +		     unsigned long combine_cruft_below_size,
     +		     struct string_list *names,
    @@ repack-cruft.c (new)
     +}
     
      ## repack.h ##
    -@@ repack.h: int write_filtered_pack(struct write_pack_opts *opts,
    +@@ repack.h: int write_filtered_pack(const struct write_pack_opts *opts,
      			struct existing_packs *existing,
      			struct string_list *names);
      
    -+int write_cruft_pack(struct write_pack_opts *opts,
    ++int write_cruft_pack(const struct write_pack_opts *opts,
     +		     const char *cruft_expiration,
     +		     unsigned long combine_cruft_below_size,
     +		     struct string_list *names,
49:  16eca91f2c = 49:  b7178565ac builtin/repack.c: clean up unused `#include`s
base-commit: 4b71b294773cc4f7fe48ec3a70079aa8783f373d
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 01/49] builtin/repack.c: avoid "the_repository" in `cmd_repack()`

Reduce builtin/repack.c's reliance on `the_repository` by using the currently-UNUSED "repo" parameter within cmd_repack().

The following commits will continue to reduce the usage of the_repository in other places within builtin/repack.c.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 31 ++++++++++++++++---------------
 1 file changed, 16 insertions(+), 15 deletions(-)
Show changes to builtin/repack.c +16 −15
diff --git a/builtin/repack.c b/builtin/repack.c
index e8730808c5..305782b2c9 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -1247,7 +1247,7 @@ static const char *find_pack_prefix(const char *packdir, const char *packtmp)
 int cmd_repack(int argc,
 	       const char **argv,
 	       const char *prefix,
-	       struct repository *repo UNUSED)
+	       struct repository *repo)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
@@ -1344,7 +1344,7 @@ int cmd_repack(int argc,
 
 	list_objects_filter_init(&po_args.filter_options);
 
-	repo_config(the_repository, repack_config, &cruft_po_args);
+	repo_config(repo, repack_config, &cruft_po_args);
 
 	argc = parse_options(argc, argv, prefix, builtin_repack_options,
 				git_repack_usage, 0);
@@ -1354,7 +1354,7 @@ int cmd_repack(int argc,
 	po_args.depth = xstrdup_or_null(opt_depth);
 	po_args.threads = xstrdup_or_null(opt_threads);
 
-	if (delete_redundant && the_repository->repository_format_precious_objects)
+	if (delete_redundant && repo->repository_format_precious_objects)
 		die(_("cannot delete packs in a precious-objects repo"));
 
 	die_for_incompatible_opt3(unpack_unreachable || (pack_everything & LOOSEN_UNREACHABLE), "-A",
@@ -1376,7 +1376,7 @@ int cmd_repack(int argc,
 		die(_(incremental_bitmap_conflict_error));
 
 	if (write_bitmaps && po_args.local &&
-	    odb_has_alternates(the_repository->objects)) {
+	    odb_has_alternates(repo->objects)) {
 		/*
 		 * When asked to do a local repack, but we have
 		 * packfiles that are inherited from an alternate, then
@@ -1391,7 +1391,8 @@ int cmd_repack(int argc,
 	if (write_midx && write_bitmaps) {
 		struct strbuf path = STRBUF_INIT;
 
-		strbuf_addf(&path, "%s/%s_XXXXXX", repo_get_object_directory(the_repository),
+		strbuf_addf(&path, "%s/%s_XXXXXX",
+			    repo_get_object_directory(repo),
 			    "bitmap-ref-tips");
 
 		refs_snapshot = xmks_tempfile(path.buf);
@@ -1400,7 +1401,7 @@ int cmd_repack(int argc,
 		strbuf_release(&path);
 	}
 
-	packdir = mkpathdup("%s/pack", repo_get_object_directory(the_repository));
+	packdir = mkpathdup("%s/pack", repo_get_object_directory(repo));
 	packtmp_name = xstrfmt(".tmp-%d-pack", (int)getpid());
 	packtmp = mkpathdup("%s/%s", packdir, packtmp_name);
 
@@ -1439,7 +1440,7 @@ int cmd_repack(int argc,
 		strvec_push(&cmd.args, "--reflog");
 		strvec_push(&cmd.args, "--indexed-objects");
 	}
-	if (repo_has_promisor_remote(the_repository))
+	if (repo_has_promisor_remote(repo))
 		strvec_push(&cmd.args, "--exclude-promisor-objects");
 	if (!write_midx) {
 		if (write_bitmaps > 0)
@@ -1535,7 +1536,7 @@ int cmd_repack(int argc,
 		 * midx_has_unknown_packs() will make the decision for
 		 * us.
 		 */
-		if (!get_multi_pack_index(the_repository->objects->sources))
+		if (!get_multi_pack_index(repo->objects->sources))
 			midx_must_contain_cruft = 1;
 	}
 
@@ -1618,9 +1619,9 @@ int cmd_repack(int argc,
 
 	string_list_sort(&names);
 
-	if (get_multi_pack_index(the_repository->objects->sources)) {
+	if (get_multi_pack_index(repo->objects->sources)) {
 		struct multi_pack_index *m =
-			get_multi_pack_index(the_repository->objects->sources);
+			get_multi_pack_index(repo->objects->sources);
 
 		ALLOC_ARRAY(midx_pack_names,
 			    m->num_packs + m->num_packs_in_base);
@@ -1631,7 +1632,7 @@ int cmd_repack(int argc,
 					xstrdup(m->pack_names[i]);
 	}
 
-	close_object_store(the_repository->objects);
+	close_object_store(repo->objects);
 
 	/*
 	 * Ok we have prepared all new packfiles.
@@ -1688,7 +1689,7 @@ int cmd_repack(int argc,
 			goto cleanup;
 	}
 
-	odb_reprepare(the_repository->objects);
+	odb_reprepare(repo->objects);
 
 	if (delete_redundant) {
 		int opts = 0;
@@ -1704,18 +1705,18 @@ int cmd_repack(int argc,
 		if (!keep_unreachable &&
 		    (!(pack_everything & LOOSEN_UNREACHABLE) ||
 		     unpack_unreachable) &&
-		    is_repository_shallow(the_repository))
+		    is_repository_shallow(repo))
 			prune_shallow(PRUNE_QUICK);
 	}
 
 	if (run_update_server_info)
-		update_server_info(the_repository, 0);
+		update_server_info(repo, 0);
 
 	if (git_env_bool(GIT_TEST_MULTI_PACK_INDEX, 0)) {
 		unsigned flags = 0;
 		if (git_env_bool(GIT_TEST_MULTI_PACK_INDEX_WRITE_INCREMENTAL, 0))
 			flags |= MIDX_WRITE_INCREMENTAL;
-		write_midx_file(the_repository->objects->sources,
+		write_midx_file(repo->objects->sources,
 				NULL, NULL, flags);
 	}
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 02/49] builtin/repack.c: avoid "the_repository" in existing packs API

There are a number of spots within builtin/repack.c which refer to "the_repository", and either make use of the "existing packs" API or otherwise have a 'struct existing_packs *' in scope.

Add a "repo" member to "struct existing_packs" and use that instead of "the_repository" in such locations.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
Show changes to builtin/repack.c +5 −3
diff --git a/builtin/repack.c b/builtin/repack.c
index 305782b2c9..7223553bed 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -126,6 +126,7 @@ static void pack_objects_args_release(struct pack_objects_args *args)
 }
 
 struct existing_packs {
+	struct repository *repo;
 	struct string_list kept_packs;
 	struct string_list non_kept_packs;
 	struct string_list cruft_packs;
@@ -265,7 +266,7 @@ static void existing_packs_release(struct existing_packs *existing)
 static void collect_pack_filenames(struct existing_packs *existing,
 				   const struct string_list *extra_keep)
 {
-	struct packfile_store *packs = the_repository->objects->packfiles;
+	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
 	struct strbuf buf = STRBUF_INIT;
 
@@ -498,7 +499,7 @@ static void init_pack_geometry(struct pack_geometry *geometry,
 			       struct existing_packs *existing,
 			       const struct pack_objects_args *args)
 {
-	struct packfile_store *packs = the_repository->objects->packfiles;
+	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
 	struct strbuf buf = STRBUF_INIT;
 
@@ -1139,7 +1140,7 @@ static int write_filtered_pack(const struct pack_objects_args *args,
 static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
 				      struct existing_packs *existing)
 {
-	struct packfile_store *packs = the_repository->objects->packfiles;
+	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
 	struct strbuf buf = STRBUF_INIT;
 	size_t i;
@@ -1405,6 +1406,7 @@ int cmd_repack(int argc,
 	packtmp_name = xstrfmt(".tmp-%d-pack", (int)getpid());
 	packtmp = mkpathdup("%s/%s", packdir, packtmp_name);
 
+	existing.repo = repo;
 	collect_pack_filenames(&existing, &keep_pack_list);
 
 	if (geometry.split_factor) {
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 03/49] builtin/repack.c: avoid "the_repository" when taking a ref snapshot

Avoid using "the_repository" in various MIDX-related ref snapshotting functions.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)
Show changes to builtin/repack.c +9 −7
diff --git a/builtin/repack.c b/builtin/repack.c
index 7223553bed..113f5fc67f 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -771,6 +771,7 @@ static int midx_has_unknown_packs(char **midx_pack_names,
 }
 
 struct midx_snapshot_ref_data {
+	struct repository *repo;
 	struct tempfile *f;
 	struct oidset seen;
 	int preferred;
@@ -784,13 +785,13 @@ static int midx_snapshot_ref_one(const char *refname UNUSED,
 	struct midx_snapshot_ref_data *data = _data;
 	struct object_id peeled;
 
-	if (!peel_iterated_oid(the_repository, oid, &peeled))
+	if (!peel_iterated_oid(data->repo, oid, &peeled))
 		oid = &peeled;
 
 	if (oidset_insert(&data->seen, oid))
 		return 0; /* already seen */
 
-	if (odb_read_object_info(the_repository->objects, oid, NULL) != OBJ_COMMIT)
+	if (odb_read_object_info(data->repo->objects, oid, NULL) != OBJ_COMMIT)
 		return 0;
 
 	fprintf(data->f->fp, "%s%s\n", data->preferred ? "+" : "",
@@ -799,11 +800,12 @@ static int midx_snapshot_ref_one(const char *refname UNUSED,
 	return 0;
 }
 
-static void midx_snapshot_refs(struct tempfile *f)
+static void midx_snapshot_refs(struct repository *repo, struct tempfile *f)
 {
 	struct midx_snapshot_ref_data data;
-	const struct string_list *preferred = bitmap_preferred_tips(the_repository);
+	const struct string_list *preferred = bitmap_preferred_tips(repo);
 
+	data.repo = repo;
 	data.f = f;
 	data.preferred = 0;
 	oidset_init(&data.seen, 0);
@@ -817,13 +819,13 @@ static void midx_snapshot_refs(struct tempfile *f)
 
 		data.preferred = 1;
 		for_each_string_list_item(item, preferred)
-			refs_for_each_ref_in(get_main_ref_store(the_repository),
+			refs_for_each_ref_in(get_main_ref_store(repo),
 					     item->string,
 					     midx_snapshot_ref_one, &data);
 		data.preferred = 0;
 	}
 
-	refs_for_each_ref(get_main_ref_store(the_repository),
+	refs_for_each_ref(get_main_ref_store(repo),
 			  midx_snapshot_ref_one, &data);
 
 	if (close_tempfile_gently(f)) {
@@ -1397,7 +1399,7 @@ int cmd_repack(int argc,
 			    "bitmap-ref-tips");
 
 		refs_snapshot = xmks_tempfile(path.buf);
-		midx_snapshot_refs(refs_snapshot);
+		midx_snapshot_refs(repo, refs_snapshot);
 
 		strbuf_release(&path);
 	}
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 04/49] builtin/repack.c: avoid "the_repository" when removing packs

The 'remove_redundant_pack()' function uses "the_repository" to obtain, and optionally remove, the repository's MIDX. Instead of relying on "the_repository", pass around a "struct repository *" parameter through its callers, and use that instead.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)
Show changes to builtin/repack.c +10 −8
diff --git a/builtin/repack.c b/builtin/repack.c
index 113f5fc67f..93802531e1 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -221,33 +221,35 @@ static void mark_packs_for_deletion(struct existing_packs *existing,
 	mark_packs_for_deletion_1(names, &existing->cruft_packs);
 }
 
-static void remove_redundant_pack(const char *dir_name, const char *base_name)
+static void remove_redundant_pack(struct repository *repo,
+				  const char *dir_name, const char *base_name)
 {
 	struct strbuf buf = STRBUF_INIT;
-	struct odb_source *source = the_repository->objects->sources;
+	struct odb_source *source = repo->objects->sources;
 	struct multi_pack_index *m = get_multi_pack_index(source);
 	strbuf_addf(&buf, "%s.pack", base_name);
 	if (m && source->local && midx_contains_pack(m, buf.buf))
-		clear_midx_file(the_repository);
+		clear_midx_file(repo);
 	strbuf_insertf(&buf, 0, "%s/", dir_name);
 	unlink_pack_path(buf.buf, 1);
 	strbuf_release(&buf);
 }
 
-static void remove_redundant_packs_1(struct string_list *packs)
+static void remove_redundant_packs_1(struct repository *repo,
+				     struct string_list *packs)
 {
 	struct string_list_item *item;
 	for_each_string_list_item(item, packs) {
 		if (!pack_is_marked_for_deletion(item))
 			continue;
-		remove_redundant_pack(packdir, item->string);
+		remove_redundant_pack(repo, packdir, item->string);
 	}
 }
 
 static void remove_redundant_existing_packs(struct existing_packs *existing)
 {
-	remove_redundant_packs_1(&existing->non_kept_packs);
-	remove_redundant_packs_1(&existing->cruft_packs);
+	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs);
+	remove_redundant_packs_1(existing->repo, &existing->cruft_packs);
 }
 
 static void existing_packs_release(struct existing_packs *existing)
@@ -685,7 +687,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
 		    (string_list_has_string(&existing->kept_packs, buf.buf)))
 			continue;
 
-		remove_redundant_pack(packdir, buf.buf);
+		remove_redundant_pack(existing->repo, packdir, buf.buf);
 	}
 
 	strbuf_release(&buf);
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 05/49] builtin/repack.c: avoid "the_repository" when repacking promisor objects

Pass a "struct repository" pointer to the 'repack_promisor_objects()' function to avoid using "the_repository".

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
Show changes to builtin/repack.c +4 −3
diff --git a/builtin/repack.c b/builtin/repack.c
index 93802531e1..4f08b57ddb 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -407,7 +407,8 @@ static int has_pack_ext(const struct generated_pack_data *data,
 	BUG("unknown pack extension: '%s'", ext);
 }
 
-static void repack_promisor_objects(const struct pack_objects_args *args,
+static void repack_promisor_objects(struct repository *repo,
+				    const struct pack_objects_args *args,
 				    struct string_list *names)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
@@ -424,7 +425,7 @@ static void repack_promisor_objects(const struct pack_objects_args *args,
 	 * {type -> existing pack order} ordering when computing deltas instead
 	 * of a {type -> size} ordering, which may produce better deltas.
 	 */
-	for_each_packed_object(the_repository, write_oid, &cmd,
+	for_each_packed_object(repo, write_oid, &cmd,
 			       FOR_EACH_OBJECT_PROMISOR_ONLY);
 
 	if (cmd.in == -1) {
@@ -1458,7 +1459,7 @@ int cmd_repack(int argc,
 		strvec_push(&cmd.args, "--delta-islands");
 
 	if (pack_everything & ALL_INTO_ONE) {
-		repack_promisor_objects(&po_args, &names);
+		repack_promisor_objects(repo, &po_args, &names);
 
 		if (has_existing_non_kept_packs(&existing) &&
 		    delete_redundant &&
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 06/49] builtin/repack.c: avoid "the_hash_algo" when deleting packs

The "mark_packs_for_deletion_1" function uses "the_hash_algo->hexsz" to isolate a pack's checksum before deleting it to avoid deleting a newly written pack having the same checksum (that is, some generated pack wound up identical to an existing pack).

Avoid this by passing down a "struct git_hash_algo" pointer, and refer to the hash algorithm through it instead.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
Show changes to builtin/repack.c +6 −4
diff --git a/builtin/repack.c b/builtin/repack.c
index 4f08b57ddb..094f5a0cc2 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -168,11 +168,12 @@ static int pack_is_retained(struct string_list_item *item)
 	return (uintptr_t)item->util & RETAIN_PACK;
 }
 
-static void mark_packs_for_deletion_1(struct string_list *names,
+static void mark_packs_for_deletion_1(const struct git_hash_algo *algop,
+				      struct string_list *names,
 				      struct string_list *list)
 {
 	struct string_list_item *item;
-	const int hexsz = the_hash_algo->hexsz;
+	const int hexsz = algop->hexsz;
 
 	for_each_string_list_item(item, list) {
 		char *sha1;
@@ -217,8 +218,9 @@ static void mark_packs_for_deletion(struct existing_packs *existing,
 				    struct string_list *names)
 
 {
-	mark_packs_for_deletion_1(names, &existing->non_kept_packs);
-	mark_packs_for_deletion_1(names, &existing->cruft_packs);
+	const struct git_hash_algo *algop = existing->repo->hash_algo;
+	mark_packs_for_deletion_1(algop, names, &existing->non_kept_packs);
+	mark_packs_for_deletion_1(algop, names, &existing->cruft_packs);
 }
 
 static void remove_redundant_pack(struct repository *repo,
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 07/49] builtin/repack.c: avoid "the_hash_algo" in `write_oid()`

In a similar spirit as the previous commit, avoid referring directly to "the_hash_algo" within builtin/repack.c::write_oid().

Unlike the previous commit, we are within a callback function, so must introduce a new struct to pass additional data through its "data" pointer.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)
Show changes to builtin/repack.c +12 −3
diff --git a/builtin/repack.c b/builtin/repack.c
index 094f5a0cc2..7d62959dc2 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -339,6 +339,11 @@ static void prepare_pack_objects(struct child_process *cmd,
 	cmd->out = -1;
 }
 
+struct write_oid_context {
+	struct child_process *cmd;
+	const struct git_hash_algo *algop;
+};
+
 /*
  * Write oid to the given struct child_process's stdin, starting it first if
  * necessary.
@@ -347,14 +352,15 @@ static int write_oid(const struct object_id *oid,
 		     struct packed_git *pack UNUSED,
 		     uint32_t pos UNUSED, void *data)
 {
-	struct child_process *cmd = data;
+	struct write_oid_context *ctx = data;
+	struct child_process *cmd = ctx->cmd;
 
 	if (cmd->in == -1) {
 		if (start_command(cmd))
 			die(_("could not start pack-objects to repack promisor objects"));
 	}
 
-	if (write_in_full(cmd->in, oid_to_hex(oid), the_hash_algo->hexsz) < 0 ||
+	if (write_in_full(cmd->in, oid_to_hex(oid), ctx->algop->hexsz) < 0 ||
 	    write_in_full(cmd->in, "\n", 1) < 0)
 		die(_("failed to feed promisor objects to pack-objects"));
 	return 0;
@@ -413,6 +419,7 @@ static void repack_promisor_objects(struct repository *repo,
 				    const struct pack_objects_args *args,
 				    struct string_list *names)
 {
+	struct write_oid_context ctx;
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	FILE *out;
 	struct strbuf line = STRBUF_INIT;
@@ -427,7 +434,9 @@ static void repack_promisor_objects(struct repository *repo,
 	 * {type -> existing pack order} ordering when computing deltas instead
 	 * of a {type -> size} ordering, which may produce better deltas.
 	 */
-	for_each_packed_object(repo, write_oid, &cmd,
+	ctx.cmd = &cmd;
+	ctx.algop = repo->hash_algo;
+	for_each_packed_object(repo, write_oid, &ctx,
 			       FOR_EACH_OBJECT_PROMISOR_ONLY);
 
 	if (cmd.in == -1) {
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 08/49] builtin/repack: avoid "the_hash_algo" in `repack_promisor_objects()`

In a similar spirit as the previous commits, avoid referring directly to "the_hash_algo" within builtin/repack.c::repack_promisor_objects().

Since there is already a repository pointer in scope, use its hash_algo value instead.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Show changes to builtin/repack.c +1 −1
diff --git a/builtin/repack.c b/builtin/repack.c
index 7d62959dc2..a7e94ed03c 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -452,7 +452,7 @@ static void repack_promisor_objects(struct repository *repo,
 		struct string_list_item *item;
 		char *promisor_name;
 
-		if (line.len != the_hash_algo->hexsz)
+		if (line.len != repo->hash_algo->hexsz)
 			die(_("repack: Expecting full hex object ID lines only from pack-objects."));
 		item = string_list_append(names, line.buf);
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 09/49] builtin/repack.c: avoid "the_hash_algo" in `finish_pack_objects_cmd()`

In a similar spirit as previous commits, avoid referring directly to "the_hash_algo" in builtin/repack.c::finish_pack_objects_cmd() and instead accept one as a parameter to the function.

Since this function has a number of callers throughout the builtin, the diff is a little noisier than previous commits. However, each hunk is limited to passing the hash_algo parameter from a repository pointer that is already in scope.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)
Show changes to builtin/repack.c +8 −5
diff --git a/builtin/repack.c b/builtin/repack.c
index a7e94ed03c..a043704aa8 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -1073,7 +1073,8 @@ static void remove_redundant_bitmaps(struct string_list *include,
 	strbuf_release(&path);
 }
 
-static int finish_pack_objects_cmd(struct child_process *cmd,
+static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
+				   struct child_process *cmd,
 				   struct string_list *names,
 				   int local)
 {
@@ -1084,7 +1085,7 @@ static int finish_pack_objects_cmd(struct child_process *cmd,
 	while (strbuf_getline_lf(&line, out) != EOF) {
 		struct string_list_item *item;
 
-		if (line.len != the_hash_algo->hexsz)
+		if (line.len != algop->hexsz)
 			die(_("repack: Expecting full hex object ID lines only "
 			      "from pack-objects."));
 		/*
@@ -1150,7 +1151,8 @@ static int write_filtered_pack(const struct pack_objects_args *args,
 		fprintf(in, "%s%s.pack\n", caret, item->string);
 	fclose(in);
 
-	return finish_pack_objects_cmd(&cmd, names, local);
+	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
+				       local);
 }
 
 static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
@@ -1247,7 +1249,8 @@ static int write_cruft_pack(const struct pack_objects_args *args,
 		fprintf(in, "%s.pack\n", item->string);
 	fclose(in);
 
-	return finish_pack_objects_cmd(&cmd, names, local);
+	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
+				       local);
 }
 
 static const char *find_pack_prefix(const char *packdir, const char *packtmp)
@@ -1534,7 +1537,7 @@ int cmd_repack(int argc,
 		fclose(in);
 	}
 
-	ret = finish_pack_objects_cmd(&cmd, &names, 1);
+	ret = finish_pack_objects_cmd(repo->hash_algo, &cmd, &names, 1);
 	if (ret)
 		goto cleanup;
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 10/49] builtin/repack.c: avoid using `hash_to_hex()` in pack geometry

In previous commits, we started passing either repository or git_hash_algo pointers around to various spots within builtin/repack.c to reduce our dependency on the_repository in the hope of undef'ing USE_THE_REPOSITORY_VARIABLE.

This commit takes us as far as we can (easily) go in that direction by removing the only use of a convenience function that only exists when USE_THE_REPOSITORY_VARIABLE is defined.

Unfortunately, the only other such function is "is_bare_repository()", which is less than straightforward to convert into, say, "repo_is_bare()", the latter of the two accepting a repository pointer.

Punt on that for now, and declare this commit as the stopping point for our efforts in the direction of undef'ing USE_THE_REPOSITORY_VARIABLE.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Show changes to builtin/repack.c +3 −1
diff --git a/builtin/repack.c b/builtin/repack.c
index a043704aa8..0d35f15b4b 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -683,12 +683,14 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
 					    struct string_list *names,
 					    struct existing_packs *existing)
 {
+	const struct git_hash_algo *algop = existing->repo->hash_algo;
 	struct strbuf buf = STRBUF_INIT;
 	uint32_t i;
 
 	for (i = 0; i < geometry->split; i++) {
 		struct packed_git *p = geometry->pack[i];
-		if (string_list_has_string(names, hash_to_hex(p->hash)))
+		if (string_list_has_string(names, hash_to_hex_algop(p->hash,
+								    algop)))
 			continue;
 
 		strbuf_reset(&buf);
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 11/49] repack: introduce new compilation unit

Over the years, builtin/repack.c has turned into a grab-bag of functionality powering the 'git repack' builtin. Among its many capabilities, it:

 - can build and spawn 'git pack-objects' commands, which in turn
   generate new packs
 - has infrastructure to manage the set of existing packs in a
   repository
 - has infrastructure to split a sequence of packs into a geometric
   progression based on object size
 - can manage both generating and combining cruft packs together
 - can write new MIDXs
to name a few.

As a result, this builtin has accumulated a lot of code, making adding new functionality difficult. In the future, 'repack' will learn how to manage a chain of incremental MIDXs, adding yet more functionality into the builtin.

As a prerequisite step, let's first move some of the functionality in the builtin into its own repack.[ch].

This will be done over the course of many steps, since there are many individual components, some of which will end up in other, yet-to-exist compilation units of their own. Some of the code movement here is also non-trivial, so performing it in individual steps will make it easier to verify.

Let's start by migrating 'struct pack_objects_args' (and the related corresponding pack_objects_args_release() function) into repack.h, and teach both the Makefile and Meson how to build the new compilation unit.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile         |  1 +
 builtin/repack.c | 25 +------------------------
 meson.build      |  1 +
 repack.c         | 11 +++++++++++
 repack.h         | 23 +++++++++++++++++++++++
 5 files changed, 37 insertions(+), 24 deletions(-)
 create mode 100644 repack.c
 create mode 100644 repack.h
Show changes to 5 files +37 −24

Makefile, builtin/repack.c, meson.build, repack.c, repack.h

diff --git a/Makefile b/Makefile
index 7ea149598d..58fc05030b 100644
--- a/Makefile
+++ b/Makefile
@@ -1249,6 +1249,7 @@ LIB_OBJS += refs/packed-backend.o
 LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
+LIB_OBJS += repack.o
 LIB_OBJS += replace-object.o
 LIB_OBJS += repo-settings.o
 LIB_OBJS += repository.o
diff --git a/builtin/repack.c b/builtin/repack.c
index 0d35f15b4b..6dfcb3327e 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -19,6 +19,7 @@
 #include "prune-packed.h"
 #include "odb.h"
 #include "promisor-remote.h"
+#include "repack.h"
 #include "shallow.h"
 #include "pack.h"
 #include "pack-bitmap.h"
@@ -53,21 +54,6 @@ static const char incremental_bitmap_conflict_error[] = N_(
 "--no-write-bitmap-index or disable the pack.writeBitmaps configuration."
 );
 
-struct pack_objects_args {
-	char *window;
-	char *window_memory;
-	char *depth;
-	char *threads;
-	unsigned long max_pack_size;
-	int no_reuse_delta;
-	int no_reuse_object;
-	int quiet;
-	int local;
-	int name_hash_version;
-	int path_walk;
-	struct list_objects_filter_options filter_options;
-};
-
 static int repack_config(const char *var, const char *value,
 			 const struct config_context *ctx, void *cb)
 {
@@ -116,15 +102,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-static void pack_objects_args_release(struct pack_objects_args *args)
-{
-	free(args->window);
-	free(args->window_memory);
-	free(args->depth);
-	free(args->threads);
-	list_objects_filter_release(&args->filter_options);
-}
-
 struct existing_packs {
 	struct repository *repo;
 	struct string_list kept_packs;
diff --git a/meson.build b/meson.build
index ec55d6a5fd..4a985ce77e 100644
--- a/meson.build
+++ b/meson.build
@@ -462,6 +462,7 @@ libgit_sources = [
   'reftable/tree.c',
   'reftable/writer.c',
   'remote.c',
+  'repack.c',
   'replace-object.c',
   'repo-settings.c',
   'repository.c',
diff --git a/repack.c b/repack.c
new file mode 100644
index 0000000000..a1f5b796fb
--- /dev/null
+++ b/repack.c
@@ -0,0 +1,11 @@
+#include "git-compat-util.h"
+#include "repack.h"
+
+void pack_objects_args_release(struct pack_objects_args *args)
+{
+	free(args->window);
+	free(args->window_memory);
+	free(args->depth);
+	free(args->threads);
+	list_objects_filter_release(&args->filter_options);
+}
diff --git a/repack.h b/repack.h
new file mode 100644
index 0000000000..421d439d5a
--- /dev/null
+++ b/repack.h
@@ -0,0 +1,23 @@
+#ifndef REPACK_H
+#define REPACK_H
+
+#include "list-objects-filter-options.h"
+
+struct pack_objects_args {
+	char *window;
+	char *window_memory;
+	char *depth;
+	char *threads;
+	unsigned long max_pack_size;
+	int no_reuse_delta;
+	int no_reuse_object;
+	int quiet;
+	int local;
+	int name_hash_version;
+	int path_walk;
+	struct list_objects_filter_options filter_options;
+};
+
+void pack_objects_args_release(struct pack_objects_args *args);
+
+#endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 12/49] builtin/repack.c: pass both pack_objects args to repack_config

A subsequent commit will remove 'delta_base_offset' as a static variable within builtin/repack.c, and reintroduce it as a member of the 'struct pack_objects_args'.

As a result, the repack_config callback will need to have both the cruft- and non-cruft 'struct pack_objects_args's in scope. Introduce a new 'struct repack_config_ctx' to allow the callee to provide both pointers to the callback.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +13 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index 6dfcb3327e..af6de8d77a 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -54,10 +54,16 @@ static const char incremental_bitmap_conflict_error[] = N_(
 "--no-write-bitmap-index or disable the pack.writeBitmaps configuration."
 );
 
+struct repack_config_ctx {
+	struct pack_objects_args *po_args;
+	struct pack_objects_args *cruft_po_args;
+};
+
 static int repack_config(const char *var, const char *value,
 			 const struct config_context *ctx, void *cb)
 {
-	struct pack_objects_args *cruft_po_args = cb;
+	struct repack_config_ctx *repack_ctx = cb;
+	struct pack_objects_args *cruft_po_args = repack_ctx->cruft_po_args;
 	if (!strcmp(var, "repack.usedeltabaseoffset")) {
 		delta_base_offset = git_config_bool(var, value);
 		return 0;
@@ -1260,6 +1266,7 @@ int cmd_repack(int argc,
 	size_t midx_pack_names_nr = 0;
 
 	/* variables to be filled by option parsing */
+	struct repack_config_ctx config_ctx;
 	int delete_redundant = 0;
 	const char *unpack_unreachable = NULL;
 	int keep_unreachable = 0;
@@ -1343,7 +1350,11 @@ int cmd_repack(int argc,
 
 	list_objects_filter_init(&po_args.filter_options);
 
-	repo_config(repo, repack_config, &cruft_po_args);
+	memset(&config_ctx, 0, sizeof(config_ctx));
+	config_ctx.po_args = &po_args;
+	config_ctx.cruft_po_args = &cruft_po_args;
+
+	repo_config(repo, repack_config, &config_ctx);
 
 	argc = parse_options(argc, argv, prefix, builtin_repack_options,
 				git_repack_usage, 0);
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:27 UTC · re: Taylor Blau · lore

[PATCH v2 13/49] repack: move 'delta_base_offset' to 'struct pack_objects_args'

The static variable 'delta_base_offset' determines whether or not we pass the "--delta-base-offset" command-line argument when spawning pack-objects as a child process. Its introduction dates back to when repack was rewritten in C, all the way back in a1bbc6c017 (repack: rewrite the shell script in C, 2013-09-15).

'struct pack_objects_args' was introduced much later on in 4571324b99 (builtin/repack.c: allow configuring cruft pack generation, 2022-05-20), but did not move the 'delta_base_offset' variable.

Since the 'delta_base_offset' is a property of an individual pack-objects command, re-introduce that variable as a member of 'struct pack_objects_args', which will enable further code movement in the subsequent commits.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 11 ++++++-----
 repack.h         |  3 +++
 2 files changed, 9 insertions(+), 5 deletions(-)
Show changes to 2 files +9 −5

builtin/repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index af6de8d77a..f4af830353 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -34,7 +34,6 @@
 #define RETAIN_PACK 2
 
 static int pack_everything;
-static int delta_base_offset = 1;
 static int pack_kept_objects = -1;
 static int write_bitmaps = -1;
 static int use_delta_islands;
@@ -63,9 +62,10 @@ static int repack_config(const char *var, const char *value,
 			 const struct config_context *ctx, void *cb)
 {
 	struct repack_config_ctx *repack_ctx = cb;
+	struct pack_objects_args *po_args = repack_ctx->po_args;
 	struct pack_objects_args *cruft_po_args = repack_ctx->cruft_po_args;
 	if (!strcmp(var, "repack.usedeltabaseoffset")) {
-		delta_base_offset = git_config_bool(var, value);
+		po_args->delta_base_offset = git_config_bool(var, value);
 		return 0;
 	}
 	if (!strcmp(var, "repack.packkeptobjects")) {
@@ -315,7 +315,7 @@ static void prepare_pack_objects(struct child_process *cmd,
 		strvec_push(&cmd->args,  "--local");
 	if (args->quiet)
 		strvec_push(&cmd->args,  "--quiet");
-	if (delta_base_offset)
+	if (args->delta_base_offset)
 		strvec_push(&cmd->args,  "--delta-base-offset");
 	strvec_push(&cmd->args, out);
 	cmd->git_cmd = 1;
@@ -1271,8 +1271,8 @@ int cmd_repack(int argc,
 	const char *unpack_unreachable = NULL;
 	int keep_unreachable = 0;
 	struct string_list keep_pack_list = STRING_LIST_INIT_NODUP;
-	struct pack_objects_args po_args = { 0 };
-	struct pack_objects_args cruft_po_args = { 0 };
+	struct pack_objects_args po_args = PACK_OBJECTS_ARGS_INIT;
+	struct pack_objects_args cruft_po_args = PACK_OBJECTS_ARGS_INIT;
 	int write_midx = 0;
 	const char *cruft_expiration = NULL;
 	const char *expire_to = NULL;
@@ -1567,6 +1567,7 @@ int cmd_repack(int argc,
 
 		cruft_po_args.local = po_args.local;
 		cruft_po_args.quiet = po_args.quiet;
+		cruft_po_args.delta_base_offset = po_args.delta_base_offset;
 
 		ret = write_cruft_pack(&cruft_po_args, packtmp, pack_prefix,
 				       cruft_expiration,
diff --git a/repack.h b/repack.h
index 421d439d5a..12632d7fec 100644
--- a/repack.h
+++ b/repack.h
@@ -15,9 +15,12 @@ struct pack_objects_args {
 	int local;
 	int name_hash_version;
 	int path_walk;
+	int delta_base_offset;
 	struct list_objects_filter_options filter_options;
 };
 
+#define PACK_OBJECTS_ARGS_INIT { .delta_base_offset = 1 }
+
 void pack_objects_args_release(struct pack_objects_args *args);
 
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 14/49] repack: remove 'prepare_pack_objects' from the builtin

Now that the 'prepare_pack_objects' function no longer refers to external, static variables, move it out to repack.h as generic functionality.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 34 ----------------------------------
 repack.c         | 35 +++++++++++++++++++++++++++++++++++
 repack.h         |  5 +++++
 3 files changed, 40 insertions(+), 34 deletions(-)
Show changes to 3 files +40 −34

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index f4af830353..ff93654cfe 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -288,40 +288,6 @@ static void collect_pack_filenames(struct existing_packs *existing,
 	strbuf_release(&buf);
 }
 
-static void prepare_pack_objects(struct child_process *cmd,
-				 const struct pack_objects_args *args,
-				 const char *out)
-{
-	strvec_push(&cmd->args, "pack-objects");
-	if (args->window)
-		strvec_pushf(&cmd->args, "--window=%s", args->window);
-	if (args->window_memory)
-		strvec_pushf(&cmd->args, "--window-memory=%s", args->window_memory);
-	if (args->depth)
-		strvec_pushf(&cmd->args, "--depth=%s", args->depth);
-	if (args->threads)
-		strvec_pushf(&cmd->args, "--threads=%s", args->threads);
-	if (args->max_pack_size)
-		strvec_pushf(&cmd->args, "--max-pack-size=%lu", args->max_pack_size);
-	if (args->no_reuse_delta)
-		strvec_pushf(&cmd->args, "--no-reuse-delta");
-	if (args->no_reuse_object)
-		strvec_pushf(&cmd->args, "--no-reuse-object");
-	if (args->name_hash_version)
-		strvec_pushf(&cmd->args, "--name-hash-version=%d", args->name_hash_version);
-	if (args->path_walk)
-		strvec_pushf(&cmd->args, "--path-walk");
-	if (args->local)
-		strvec_push(&cmd->args,  "--local");
-	if (args->quiet)
-		strvec_push(&cmd->args,  "--quiet");
-	if (args->delta_base_offset)
-		strvec_push(&cmd->args,  "--delta-base-offset");
-	strvec_push(&cmd->args, out);
-	cmd->git_cmd = 1;
-	cmd->out = -1;
-}
-
 struct write_oid_context {
 	struct child_process *cmd;
 	const struct git_hash_algo *algop;
diff --git a/repack.c b/repack.c
index a1f5b796fb..91b6e1cc09 100644
--- a/repack.c
+++ b/repack.c
@@ -1,5 +1,40 @@
 #include "git-compat-util.h"
 #include "repack.h"
+#include "run-command.h"
+
+void prepare_pack_objects(struct child_process *cmd,
+			  const struct pack_objects_args *args,
+			  const char *out)
+{
+	strvec_push(&cmd->args, "pack-objects");
+	if (args->window)
+		strvec_pushf(&cmd->args, "--window=%s", args->window);
+	if (args->window_memory)
+		strvec_pushf(&cmd->args, "--window-memory=%s", args->window_memory);
+	if (args->depth)
+		strvec_pushf(&cmd->args, "--depth=%s", args->depth);
+	if (args->threads)
+		strvec_pushf(&cmd->args, "--threads=%s", args->threads);
+	if (args->max_pack_size)
+		strvec_pushf(&cmd->args, "--max-pack-size=%lu", args->max_pack_size);
+	if (args->no_reuse_delta)
+		strvec_pushf(&cmd->args, "--no-reuse-delta");
+	if (args->no_reuse_object)
+		strvec_pushf(&cmd->args, "--no-reuse-object");
+	if (args->name_hash_version)
+		strvec_pushf(&cmd->args, "--name-hash-version=%d", args->name_hash_version);
+	if (args->path_walk)
+		strvec_pushf(&cmd->args, "--path-walk");
+	if (args->local)
+		strvec_push(&cmd->args,  "--local");
+	if (args->quiet)
+		strvec_push(&cmd->args,  "--quiet");
+	if (args->delta_base_offset)
+		strvec_push(&cmd->args,  "--delta-base-offset");
+	strvec_push(&cmd->args, out);
+	cmd->git_cmd = 1;
+	cmd->out = -1;
+}
 
 void pack_objects_args_release(struct pack_objects_args *args)
 {
diff --git a/repack.h b/repack.h
index 12632d7fec..3f7ec20735 100644
--- a/repack.h
+++ b/repack.h
@@ -21,6 +21,11 @@ struct pack_objects_args {
 
 #define PACK_OBJECTS_ARGS_INIT { .delta_base_offset = 1 }
 
+struct child_process;
+
+void prepare_pack_objects(struct child_process *cmd,
+			  const struct pack_objects_args *args,
+			  const char *out);
 void pack_objects_args_release(struct pack_objects_args *args);
 
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 15/49] builtin/repack.c: rename many 'struct existing_packs' functions

Rename many of the 'struct existing_packs'-related functions according to the convention introduced in and described by 541204aabe (Documentation: document naming schema for structs and their functions, 2024-07-30).

Note that some functions which operate over an individual entry in the list of existing packs are prefixed with "existing_pack_" instead of the plural form.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 66 +++++++++++++++++++++++++-----------------------
 1 file changed, 34 insertions(+), 32 deletions(-)
Show changes to builtin/repack.c +34 −32
diff --git a/builtin/repack.c b/builtin/repack.c
index ff93654cfe..f82e6c3930 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -121,39 +121,39 @@ struct existing_packs {
 	.cruft_packs = STRING_LIST_INIT_DUP, \
 }
 
-static int has_existing_non_kept_packs(const struct existing_packs *existing)
+static int existing_packs_has_non_kept(const struct existing_packs *existing)
 {
 	return existing->non_kept_packs.nr || existing->cruft_packs.nr;
 }
 
-static void pack_mark_for_deletion(struct string_list_item *item)
+static void existing_pack_mark_for_deletion(struct string_list_item *item)
 {
 	item->util = (void*)((uintptr_t)item->util | DELETE_PACK);
 }
 
-static void pack_unmark_for_deletion(struct string_list_item *item)
+static void existing_pack_unmark_for_deletion(struct string_list_item *item)
 {
 	item->util = (void*)((uintptr_t)item->util & ~DELETE_PACK);
 }
 
-static int pack_is_marked_for_deletion(struct string_list_item *item)
+static int existing_pack_is_marked_for_deletion(struct string_list_item *item)
 {
 	return (uintptr_t)item->util & DELETE_PACK;
 }
 
-static void pack_mark_retained(struct string_list_item *item)
+static void existing_packs_mark_retained(struct string_list_item *item)
 {
 	item->util = (void*)((uintptr_t)item->util | RETAIN_PACK);
 }
 
-static int pack_is_retained(struct string_list_item *item)
+static int existing_pack_is_retained(struct string_list_item *item)
 {
 	return (uintptr_t)item->util & RETAIN_PACK;
 }
 
-static void mark_packs_for_deletion_1(const struct git_hash_algo *algop,
-				      struct string_list *names,
-				      struct string_list *list)
+static void existing_packs_mark_for_deletion_1(const struct git_hash_algo *algop,
+					       struct string_list *names,
+					       struct string_list *list)
 {
 	struct string_list_item *item;
 	const int hexsz = algop->hexsz;
@@ -165,8 +165,8 @@ static void mark_packs_for_deletion_1(const struct git_hash_algo *algop,
 			continue;
 		sha1 = item->string + len - hexsz;
 
-		if (pack_is_retained(item)) {
-			pack_unmark_for_deletion(item);
+		if (existing_pack_is_retained(item)) {
+			existing_pack_unmark_for_deletion(item);
 		} else if (!string_list_has_string(names, sha1)) {
 			/*
 			 * Mark this pack for deletion, which ensures
@@ -175,13 +175,13 @@ static void mark_packs_for_deletion_1(const struct git_hash_algo *algop,
 			 * will actually delete this pack (if `-d` was
 			 * given).
 			 */
-			pack_mark_for_deletion(item);
+			existing_pack_mark_for_deletion(item);
 		}
 	}
 }
 
-static void retain_cruft_pack(struct existing_packs *existing,
-			      struct packed_git *cruft)
+static void existing_packs_retain_cruft(struct existing_packs *existing,
+					struct packed_git *cruft)
 {
 	struct strbuf buf = STRBUF_INIT;
 	struct string_list_item *item;
@@ -193,17 +193,19 @@ static void retain_cruft_pack(struct existing_packs *existing,
 	if (!item)
 		BUG("could not find cruft pack '%s'", pack_basename(cruft));
 
-	pack_mark_retained(item);
+	existing_packs_mark_retained(item);
 	strbuf_release(&buf);
 }
 
-static void mark_packs_for_deletion(struct existing_packs *existing,
-				    struct string_list *names)
+static void existing_packs_mark_for_deletion(struct existing_packs *existing,
+					     struct string_list *names)
 
 {
 	const struct git_hash_algo *algop = existing->repo->hash_algo;
-	mark_packs_for_deletion_1(algop, names, &existing->non_kept_packs);
-	mark_packs_for_deletion_1(algop, names, &existing->cruft_packs);
+	existing_packs_mark_for_deletion_1(algop, names,
+					   &existing->non_kept_packs);
+	existing_packs_mark_for_deletion_1(algop, names,
+					   &existing->cruft_packs);
 }
 
 static void remove_redundant_pack(struct repository *repo,
@@ -225,13 +227,13 @@ static void remove_redundant_packs_1(struct repository *repo,
 {
 	struct string_list_item *item;
 	for_each_string_list_item(item, packs) {
-		if (!pack_is_marked_for_deletion(item))
+		if (!existing_pack_is_marked_for_deletion(item))
 			continue;
 		remove_redundant_pack(repo, packdir, item->string);
 	}
 }
 
-static void remove_redundant_existing_packs(struct existing_packs *existing)
+static void existing_packs_remove_redundant(struct existing_packs *existing)
 {
 	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs);
 	remove_redundant_packs_1(existing->repo, &existing->cruft_packs);
@@ -250,7 +252,7 @@ static void existing_packs_release(struct existing_packs *existing)
  * .keep file or not.  Packs without a .keep file are not to be kept
  * if we are going to pack everything into one file.
  */
-static void collect_pack_filenames(struct existing_packs *existing,
+static void existing_packs_collect(struct existing_packs *existing,
 				   const struct string_list *extra_keep)
 {
 	struct packfile_store *packs = existing->repo->objects->packfiles;
@@ -721,7 +723,7 @@ static int midx_has_unknown_packs(char **midx_pack_names,
 
 			item = string_list_lookup(&existing->non_kept_packs,
 						  pack_name);
-			if (item && !pack_is_marked_for_deletion(item))
+			if (item && !existing_pack_is_marked_for_deletion(item))
 				continue;
 		}
 
@@ -851,7 +853,7 @@ static void midx_included_packs(struct string_list *include,
 		}
 	} else {
 		for_each_string_list_item(item, &existing->non_kept_packs) {
-			if (pack_is_marked_for_deletion(item))
+			if (existing_pack_is_marked_for_deletion(item))
 				continue;
 
 			strbuf_reset(&buf);
@@ -888,10 +890,10 @@ static void midx_included_packs(struct string_list *include,
 			 * --geometric case, but doing so is unnecessary
 			 *  since no packs are marked as pending
 			 *  deletion (since we only call
-			 *  `mark_packs_for_deletion()` when doing an
-			 *  all-into-one repack).
+			 *  `existing_packs_mark_for_deletion()` when
+			 *  doing an all-into-one repack).
 			 */
-			if (pack_is_marked_for_deletion(item))
+			if (existing_pack_is_marked_for_deletion(item))
 				continue;
 
 			strbuf_reset(&buf);
@@ -1128,7 +1130,7 @@ static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
 		if (p->pack_size < combine_cruft_below_size) {
 			fprintf(in, "-%s\n", pack_basename(p));
 		} else {
-			retain_cruft_pack(existing, p);
+			existing_packs_retain_cruft(existing, p);
 			fprintf(in, "%s\n", pack_basename(p));
 		}
 	}
@@ -1382,7 +1384,7 @@ int cmd_repack(int argc,
 	packtmp = mkpathdup("%s/%s", packdir, packtmp_name);
 
 	existing.repo = repo;
-	collect_pack_filenames(&existing, &keep_pack_list);
+	existing_packs_collect(&existing, &keep_pack_list);
 
 	if (geometry.split_factor) {
 		if (pack_everything)
@@ -1431,7 +1433,7 @@ int cmd_repack(int argc,
 	if (pack_everything & ALL_INTO_ONE) {
 		repack_promisor_objects(repo, &po_args, &names);
 
-		if (has_existing_non_kept_packs(&existing) &&
+		if (existing_packs_has_non_kept(&existing) &&
 		    delete_redundant &&
 		    !(pack_everything & PACK_CRUFT)) {
 			for_each_string_list_item(item, &names) {
@@ -1647,7 +1649,7 @@ int cmd_repack(int argc,
 	/* End of pack replacement. */
 
 	if (delete_redundant && pack_everything & ALL_INTO_ONE)
-		mark_packs_for_deletion(&existing, &names);
+		existing_packs_mark_for_deletion(&existing, &names);
 
 	if (write_midx) {
 		struct string_list include = STRING_LIST_INIT_DUP;
@@ -1671,7 +1673,7 @@ int cmd_repack(int argc,
 
 	if (delete_redundant) {
 		int opts = 0;
-		remove_redundant_existing_packs(&existing);
+		existing_packs_remove_redundant(&existing);
 
 		if (geometry.split_factor)
 			geometry_remove_redundant_packs(&geometry, &names,
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 16/49] repack: remove 'remove_redundant_pack' from the builtin

Extract "remove_redundant_pack()" as generic repack-related functionality by moving its implementation to the repack.[ch] compilation unit.

This is a prerequisite to moving the "existing_packs" API, which is one of the callers of this function. (The remaining caller in the pack geometry code will eventually move to its own compilation unit as well, and will likewise rely on this function.)

While moving it over, prefix the function name with "repack_" to indicate that it belongs to the repack-subsystem.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 18 ++----------------
 repack.c         | 18 ++++++++++++++++++
 repack.h         |  3 +++
 3 files changed, 23 insertions(+), 16 deletions(-)
Show changes to 3 files +23 −16

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index f82e6c3930..31137cf711 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -208,20 +208,6 @@ static void existing_packs_mark_for_deletion(struct existing_packs *existing,
 					   &existing->cruft_packs);
 }
 
-static void remove_redundant_pack(struct repository *repo,
-				  const char *dir_name, const char *base_name)
-{
-	struct strbuf buf = STRBUF_INIT;
-	struct odb_source *source = repo->objects->sources;
-	struct multi_pack_index *m = get_multi_pack_index(source);
-	strbuf_addf(&buf, "%s.pack", base_name);
-	if (m && source->local && midx_contains_pack(m, buf.buf))
-		clear_midx_file(repo);
-	strbuf_insertf(&buf, 0, "%s/", dir_name);
-	unlink_pack_path(buf.buf, 1);
-	strbuf_release(&buf);
-}
-
 static void remove_redundant_packs_1(struct repository *repo,
 				     struct string_list *packs)
 {
@@ -229,7 +215,7 @@ static void remove_redundant_packs_1(struct repository *repo,
 	for_each_string_list_item(item, packs) {
 		if (!existing_pack_is_marked_for_deletion(item))
 			continue;
-		remove_redundant_pack(repo, packdir, item->string);
+		repack_remove_redundant_pack(repo, packdir, item->string);
 	}
 }
 
@@ -652,7 +638,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
 		    (string_list_has_string(&existing->kept_packs, buf.buf)))
 			continue;
 
-		remove_redundant_pack(existing->repo, packdir, buf.buf);
+		repack_remove_redundant_pack(existing->repo, packdir, buf.buf);
 	}
 
 	strbuf_release(&buf);
diff --git a/repack.c b/repack.c
index 91b6e1cc09..3aaa351b5b 100644
--- a/repack.c
+++ b/repack.c
@@ -1,5 +1,9 @@
 #include "git-compat-util.h"
+#include "midx.h"
+#include "odb.h"
+#include "packfile.h"
 #include "repack.h"
+#include "repository.h"
 #include "run-command.h"
 
 void prepare_pack_objects(struct child_process *cmd,
@@ -44,3 +48,17 @@ void pack_objects_args_release(struct pack_objects_args *args)
 	free(args->threads);
 	list_objects_filter_release(&args->filter_options);
 }
+
+void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
+				  const char *base_name)
+{
+	struct strbuf buf = STRBUF_INIT;
+	struct odb_source *source = repo->objects->sources;
+	struct multi_pack_index *m = get_multi_pack_index(source);
+	strbuf_addf(&buf, "%s.pack", base_name);
+	if (m && source->local && midx_contains_pack(m, buf.buf))
+		clear_midx_file(repo);
+	strbuf_insertf(&buf, 0, "%s/", dir_name);
+	unlink_pack_path(buf.buf, 1);
+	strbuf_release(&buf);
+}
diff --git a/repack.h b/repack.h
index 3f7ec20735..a62bfa2ff9 100644
--- a/repack.h
+++ b/repack.h
@@ -28,4 +28,7 @@ void prepare_pack_objects(struct child_process *cmd,
 			  const char *out);
 void pack_objects_args_release(struct pack_objects_args *args);
 
+void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
+				  const char *base_name);
+
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 17/49] builtin/repack.c: pass "packdir" when removing packs

builtin/repack.c defines a static "packdir" to instruct pack-objects on where to write any new packfiles. This is also the directory scanned when removing any packfiles which were made redundant by the latest repack.

Prepare to move the "existing_packs_remove_redundant" function to its own compilation unit by passing in this information as a parameter to that function.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
Show changes to builtin/repack.c +9 −5
diff --git a/builtin/repack.c b/builtin/repack.c
index 31137cf711..c5a88eda12 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -209,7 +209,8 @@ static void existing_packs_mark_for_deletion(struct existing_packs *existing,
 }
 
 static void remove_redundant_packs_1(struct repository *repo,
-				     struct string_list *packs)
+				     struct string_list *packs,
+				     const char *packdir)
 {
 	struct string_list_item *item;
 	for_each_string_list_item(item, packs) {
@@ -219,10 +220,13 @@ static void remove_redundant_packs_1(struct repository *repo,
 	}
 }
 
-static void existing_packs_remove_redundant(struct existing_packs *existing)
+static void existing_packs_remove_redundant(struct existing_packs *existing,
+					    const char *packdir)
 {
-	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs);
-	remove_redundant_packs_1(existing->repo, &existing->cruft_packs);
+	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs,
+				 packdir);
+	remove_redundant_packs_1(existing->repo, &existing->cruft_packs,
+				 packdir);
 }
 
 static void existing_packs_release(struct existing_packs *existing)
@@ -1659,7 +1663,7 @@ int cmd_repack(int argc,
 
 	if (delete_redundant) {
 		int opts = 0;
-		existing_packs_remove_redundant(&existing);
+		existing_packs_remove_redundant(&existing, packdir);
 
 		if (geometry.split_factor)
 			geometry_remove_redundant_packs(&geometry, &names,
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 18/49] builtin/repack.c: avoid unnecessary numeric casts in existing_packs

There are a couple of spots that cause warnings within the existing_packs API without DISABLE_SIGN_COMPARE_WARNINGS under DEVELOPER=1 mode.

In both cases, we have int values that are being compared against size_t ones. Neither of these two cases are incorrect, and the cast is completely OK in practice. But both are unnecessary, since:

 - in existing_packs_mark_for_deletion_1(), 'hexsz' should be defined as
   a size_t anyway, since algop->hexsz is.
 - in existing_packs_collect(), 'i' should be defined as a size_t since
   it is counting up to the value of a string_list's 'nr' field.

(This patch is a little bit of noise, but I would rather see us squelch these warnings ahead of moving the existing_packs API into a separate compilation unit to avoid having to define DISABLE_SIGN_COMPARE_WARNINGS in repack.c.)

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +2 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index c5a88eda12..e13943b637 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -156,7 +156,7 @@ static void existing_packs_mark_for_deletion_1(const struct git_hash_algo *algop
 					       struct string_list *list)
 {
 	struct string_list_item *item;
-	const int hexsz = algop->hexsz;
+	const size_t hexsz = algop->hexsz;
 
 	for_each_string_list_item(item, list) {
 		char *sha1;
@@ -250,7 +250,7 @@ static void existing_packs_collect(struct existing_packs *existing,
 	struct strbuf buf = STRBUF_INIT;
 
 	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
-		int i;
+		size_t i;
 		const char *base;
 
 		if (!p->pack_local)
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 19/49] repack: remove 'existing_packs' API from the builtin

The repack builtin defines an API for keeping track of which packs were found in the repository at the beginning of the repack operation. This is used to classify what state a pack was in (kept, non-kept, or cruft), and is also used to mark which packs to delete (or keep) at the end of a repack operation.

Now that the prerequisite refactoring is complete, this API is isolated enough that it can be moved out to repack.[ch] and removed from the builtin entirely.

As a result, some of its functions become static within repack.c, cleaning up the visible API.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 173 -----------------------------------------------
 repack.c         | 157 ++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  35 ++++++++++
 3 files changed, 192 insertions(+), 173 deletions(-)
Show changes to 3 files +192 −173

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index e13943b637..a168c88791 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -3,7 +3,6 @@
 
 #include "builtin.h"
 #include "config.h"
-#include "dir.h"
 #include "environment.h"
 #include "gettext.h"
 #include "hex.h"
@@ -108,178 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-struct existing_packs {
-	struct repository *repo;
-	struct string_list kept_packs;
-	struct string_list non_kept_packs;
-	struct string_list cruft_packs;
-};
-
-#define EXISTING_PACKS_INIT { \
-	.kept_packs = STRING_LIST_INIT_DUP, \
-	.non_kept_packs = STRING_LIST_INIT_DUP, \
-	.cruft_packs = STRING_LIST_INIT_DUP, \
-}
-
-static int existing_packs_has_non_kept(const struct existing_packs *existing)
-{
-	return existing->non_kept_packs.nr || existing->cruft_packs.nr;
-}
-
-static void existing_pack_mark_for_deletion(struct string_list_item *item)
-{
-	item->util = (void*)((uintptr_t)item->util | DELETE_PACK);
-}
-
-static void existing_pack_unmark_for_deletion(struct string_list_item *item)
-{
-	item->util = (void*)((uintptr_t)item->util & ~DELETE_PACK);
-}
-
-static int existing_pack_is_marked_for_deletion(struct string_list_item *item)
-{
-	return (uintptr_t)item->util & DELETE_PACK;
-}
-
-static void existing_packs_mark_retained(struct string_list_item *item)
-{
-	item->util = (void*)((uintptr_t)item->util | RETAIN_PACK);
-}
-
-static int existing_pack_is_retained(struct string_list_item *item)
-{
-	return (uintptr_t)item->util & RETAIN_PACK;
-}
-
-static void existing_packs_mark_for_deletion_1(const struct git_hash_algo *algop,
-					       struct string_list *names,
-					       struct string_list *list)
-{
-	struct string_list_item *item;
-	const size_t hexsz = algop->hexsz;
-
-	for_each_string_list_item(item, list) {
-		char *sha1;
-		size_t len = strlen(item->string);
-		if (len < hexsz)
-			continue;
-		sha1 = item->string + len - hexsz;
-
-		if (existing_pack_is_retained(item)) {
-			existing_pack_unmark_for_deletion(item);
-		} else if (!string_list_has_string(names, sha1)) {
-			/*
-			 * Mark this pack for deletion, which ensures
-			 * that this pack won't be included in a MIDX
-			 * (if `--write-midx` was given) and that we
-			 * will actually delete this pack (if `-d` was
-			 * given).
-			 */
-			existing_pack_mark_for_deletion(item);
-		}
-	}
-}
-
-static void existing_packs_retain_cruft(struct existing_packs *existing,
-					struct packed_git *cruft)
-{
-	struct strbuf buf = STRBUF_INIT;
-	struct string_list_item *item;
-
-	strbuf_addstr(&buf, pack_basename(cruft));
-	strbuf_strip_suffix(&buf, ".pack");
-
-	item = string_list_lookup(&existing->cruft_packs, buf.buf);
-	if (!item)
-		BUG("could not find cruft pack '%s'", pack_basename(cruft));
-
-	existing_packs_mark_retained(item);
-	strbuf_release(&buf);
-}
-
-static void existing_packs_mark_for_deletion(struct existing_packs *existing,
-					     struct string_list *names)
-
-{
-	const struct git_hash_algo *algop = existing->repo->hash_algo;
-	existing_packs_mark_for_deletion_1(algop, names,
-					   &existing->non_kept_packs);
-	existing_packs_mark_for_deletion_1(algop, names,
-					   &existing->cruft_packs);
-}
-
-static void remove_redundant_packs_1(struct repository *repo,
-				     struct string_list *packs,
-				     const char *packdir)
-{
-	struct string_list_item *item;
-	for_each_string_list_item(item, packs) {
-		if (!existing_pack_is_marked_for_deletion(item))
-			continue;
-		repack_remove_redundant_pack(repo, packdir, item->string);
-	}
-}
-
-static void existing_packs_remove_redundant(struct existing_packs *existing,
-					    const char *packdir)
-{
-	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs,
-				 packdir);
-	remove_redundant_packs_1(existing->repo, &existing->cruft_packs,
-				 packdir);
-}
-
-static void existing_packs_release(struct existing_packs *existing)
-{
-	string_list_clear(&existing->kept_packs, 0);
-	string_list_clear(&existing->non_kept_packs, 0);
-	string_list_clear(&existing->cruft_packs, 0);
-}
-
-/*
- * Adds all packs hex strings (pack-$HASH) to either packs->non_kept
- * or packs->kept based on whether each pack has a corresponding
- * .keep file or not.  Packs without a .keep file are not to be kept
- * if we are going to pack everything into one file.
- */
-static void existing_packs_collect(struct existing_packs *existing,
-				   const struct string_list *extra_keep)
-{
-	struct packfile_store *packs = existing->repo->objects->packfiles;
-	struct packed_git *p;
-	struct strbuf buf = STRBUF_INIT;
-
-	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
-		size_t i;
-		const char *base;
-
-		if (!p->pack_local)
-			continue;
-
-		base = pack_basename(p);
-
-		for (i = 0; i < extra_keep->nr; i++)
-			if (!fspathcmp(base, extra_keep->items[i].string))
-				break;
-
-		strbuf_reset(&buf);
-		strbuf_addstr(&buf, base);
-		strbuf_strip_suffix(&buf, ".pack");
-
-		if ((extra_keep->nr > 0 && i < extra_keep->nr) || p->pack_keep)
-			string_list_append(&existing->kept_packs, buf.buf);
-		else if (p->is_cruft)
-			string_list_append(&existing->cruft_packs, buf.buf);
-		else
-			string_list_append(&existing->non_kept_packs, buf.buf);
-	}
-
-	string_list_sort(&existing->kept_packs);
-	string_list_sort(&existing->non_kept_packs);
-	string_list_sort(&existing->cruft_packs);
-	strbuf_release(&buf);
-}
-
 struct write_oid_context {
 	struct child_process *cmd;
 	const struct git_hash_algo *algop;
diff --git a/repack.c b/repack.c
index 3aaa351b5b..9182e1c50b 100644
--- a/repack.c
+++ b/repack.c
@@ -1,4 +1,5 @@
 #include "git-compat-util.h"
+#include "dir.h"
 #include "midx.h"
 #include "odb.h"
 #include "packfile.h"
@@ -62,3 +63,159 @@ void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 	unlink_pack_path(buf.buf, 1);
 	strbuf_release(&buf);
 }
+
+#define DELETE_PACK 1
+#define RETAIN_PACK 2
+
+void existing_packs_collect(struct existing_packs *existing,
+			    const struct string_list *extra_keep)
+{
+	struct packfile_store *packs = existing->repo->objects->packfiles;
+	struct packed_git *p;
+	struct strbuf buf = STRBUF_INIT;
+
+	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
+		size_t i;
+		const char *base;
+
+		if (!p->pack_local)
+			continue;
+
+		base = pack_basename(p);
+
+		for (i = 0; i < extra_keep->nr; i++)
+			if (!fspathcmp(base, extra_keep->items[i].string))
+				break;
+
+		strbuf_reset(&buf);
+		strbuf_addstr(&buf, base);
+		strbuf_strip_suffix(&buf, ".pack");
+
+		if ((extra_keep->nr > 0 && i < extra_keep->nr) || p->pack_keep)
+			string_list_append(&existing->kept_packs, buf.buf);
+		else if (p->is_cruft)
+			string_list_append(&existing->cruft_packs, buf.buf);
+		else
+			string_list_append(&existing->non_kept_packs, buf.buf);
+	}
+
+	string_list_sort(&existing->kept_packs);
+	string_list_sort(&existing->non_kept_packs);
+	string_list_sort(&existing->cruft_packs);
+	strbuf_release(&buf);
+}
+
+int existing_packs_has_non_kept(const struct existing_packs *existing)
+{
+	return existing->non_kept_packs.nr || existing->cruft_packs.nr;
+}
+
+static void existing_pack_mark_for_deletion(struct string_list_item *item)
+{
+	item->util = (void*)((uintptr_t)item->util | DELETE_PACK);
+}
+
+static void existing_pack_unmark_for_deletion(struct string_list_item *item)
+{
+	item->util = (void*)((uintptr_t)item->util & ~DELETE_PACK);
+}
+
+int existing_pack_is_marked_for_deletion(struct string_list_item *item)
+{
+	return (uintptr_t)item->util & DELETE_PACK;
+}
+
+static void existing_packs_mark_retained(struct string_list_item *item)
+{
+	item->util = (void*)((uintptr_t)item->util | RETAIN_PACK);
+}
+
+static int existing_pack_is_retained(struct string_list_item *item)
+{
+	return (uintptr_t)item->util & RETAIN_PACK;
+}
+
+static void existing_packs_mark_for_deletion_1(const struct git_hash_algo *algop,
+					       struct string_list *names,
+					       struct string_list *list)
+{
+	struct string_list_item *item;
+	const size_t hexsz = algop->hexsz;
+
+	for_each_string_list_item(item, list) {
+		char *sha1;
+		size_t len = strlen(item->string);
+		if (len < hexsz)
+			continue;
+		sha1 = item->string + len - hexsz;
+
+		if (existing_pack_is_retained(item)) {
+			existing_pack_unmark_for_deletion(item);
+		} else if (!string_list_has_string(names, sha1)) {
+			/*
+			 * Mark this pack for deletion, which ensures
+			 * that this pack won't be included in a MIDX
+			 * (if `--write-midx` was given) and that we
+			 * will actually delete this pack (if `-d` was
+			 * given).
+			 */
+			existing_pack_mark_for_deletion(item);
+		}
+	}
+}
+
+void existing_packs_retain_cruft(struct existing_packs *existing,
+				 struct packed_git *cruft)
+{
+	struct strbuf buf = STRBUF_INIT;
+	struct string_list_item *item;
+
+	strbuf_addstr(&buf, pack_basename(cruft));
+	strbuf_strip_suffix(&buf, ".pack");
+
+	item = string_list_lookup(&existing->cruft_packs, buf.buf);
+	if (!item)
+		BUG("could not find cruft pack '%s'", pack_basename(cruft));
+
+	existing_packs_mark_retained(item);
+	strbuf_release(&buf);
+}
+
+void existing_packs_mark_for_deletion(struct existing_packs *existing,
+				      struct string_list *names)
+
+{
+	const struct git_hash_algo *algop = existing->repo->hash_algo;
+	existing_packs_mark_for_deletion_1(algop, names,
+					   &existing->non_kept_packs);
+	existing_packs_mark_for_deletion_1(algop, names,
+					   &existing->cruft_packs);
+}
+
+static void remove_redundant_packs_1(struct repository *repo,
+				     struct string_list *packs,
+				     const char *packdir)
+{
+	struct string_list_item *item;
+	for_each_string_list_item(item, packs) {
+		if (!existing_pack_is_marked_for_deletion(item))
+			continue;
+		repack_remove_redundant_pack(repo, packdir, item->string);
+	}
+}
+
+void existing_packs_remove_redundant(struct existing_packs *existing,
+				     const char *packdir)
+{
+	remove_redundant_packs_1(existing->repo, &existing->non_kept_packs,
+				 packdir);
+	remove_redundant_packs_1(existing->repo, &existing->cruft_packs,
+				 packdir);
+}
+
+void existing_packs_release(struct existing_packs *existing)
+{
+	string_list_clear(&existing->kept_packs, 0);
+	string_list_clear(&existing->non_kept_packs, 0);
+	string_list_clear(&existing->cruft_packs, 0);
+}
diff --git a/repack.h b/repack.h
index a62bfa2ff9..19796e2243 100644
--- a/repack.h
+++ b/repack.h
@@ -2,6 +2,7 @@
 #define REPACK_H
 
 #include "list-objects-filter-options.h"
+#include "string-list.h"
 
 struct pack_objects_args {
 	char *window;
@@ -31,4 +32,38 @@ void pack_objects_args_release(struct pack_objects_args *args);
 void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 				  const char *base_name);
 
+struct repository;
+struct packed_git;
+
+struct existing_packs {
+	struct repository *repo;
+	struct string_list kept_packs;
+	struct string_list non_kept_packs;
+	struct string_list cruft_packs;
+};
+
+#define EXISTING_PACKS_INIT { \
+	.kept_packs = STRING_LIST_INIT_DUP, \
+	.non_kept_packs = STRING_LIST_INIT_DUP, \
+	.cruft_packs = STRING_LIST_INIT_DUP, \
+}
+
+/*
+ * Adds all packs hex strings (pack-$HASH) to either packs->non_kept
+ * or packs->kept based on whether each pack has a corresponding
+ * .keep file or not.  Packs without a .keep file are not to be kept
+ * if we are going to pack everything into one file.
+ */
+void existing_packs_collect(struct existing_packs *existing,
+			    const struct string_list *extra_keep);
+int existing_packs_has_non_kept(const struct existing_packs *existing);
+int existing_pack_is_marked_for_deletion(struct string_list_item *item);
+void existing_packs_retain_cruft(struct existing_packs *existing,
+				 struct packed_git *cruft);
+void existing_packs_mark_for_deletion(struct existing_packs *existing,
+				      struct string_list *names);
+void existing_packs_remove_redundant(struct existing_packs *existing,
+				     const char *packdir);
+void existing_packs_release(struct existing_packs *existing);
+
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 20/49] builtin/repack.c: rename "struct generated_pack_data"

The name "generated_pack_data" is somewhat redundant, since the contents of the struct *is* the data associated with the generated pack.

Rename the structure to just "generated_pack", resulting in less awkward function names, like "generated_pack_has_ext()" which is preferable to "generated_pack_data_has_ext()".

Rename a few related functions to align with the convention that functions to do with a struct "S" should be prefixed with "S_".

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 32 ++++++++++++++++----------------
 1 file changed, 16 insertions(+), 16 deletions(-)
Show changes to builtin/repack.c +16 −16
diff --git a/builtin/repack.c b/builtin/repack.c
index a168c88791..a4d80b6b04 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -146,15 +146,15 @@ static struct {
 	{".idx"},
 };
 
-struct generated_pack_data {
+struct generated_pack {
 	struct tempfile *tempfiles[ARRAY_SIZE(exts)];
 };
 
-static struct generated_pack_data *populate_pack_exts(const char *name)
+static struct generated_pack *generated_pack_populate(const char *name)
 {
 	struct stat statbuf;
 	struct strbuf path = STRBUF_INIT;
-	struct generated_pack_data *data = xcalloc(1, sizeof(*data));
+	struct generated_pack *pack = xcalloc(1, sizeof(*pack));
 	int i;
 
 	for (i = 0; i < ARRAY_SIZE(exts); i++) {
@@ -164,21 +164,21 @@ static struct generated_pack_data *populate_pack_exts(const char *name)
 		if (stat(path.buf, &statbuf))
 			continue;
 
-		data->tempfiles[i] = register_tempfile(path.buf);
+		pack->tempfiles[i] = register_tempfile(path.buf);
 	}
 
 	strbuf_release(&path);
-	return data;
+	return pack;
 }
 
-static int has_pack_ext(const struct generated_pack_data *data,
-			const char *ext)
+static int generated_pack_has_ext(const struct generated_pack *pack,
+				  const char *ext)
 {
 	int i;
 	for (i = 0; i < ARRAY_SIZE(exts); i++) {
 		if (strcmp(exts[i].name, ext))
 			continue;
-		return !!data->tempfiles[i];
+		return !!pack->tempfiles[i];
 	}
 	BUG("unknown pack extension: '%s'", ext);
 }
@@ -239,7 +239,7 @@ static void repack_promisor_objects(struct repository *repo,
 					  line.buf);
 		write_promisor_file(promisor_name, NULL, 0);
 
-		item->util = populate_pack_exts(item->string);
+		item->util = generated_pack_populate(item->string);
 
 		free(promisor_name);
 	}
@@ -780,8 +780,8 @@ static int write_midx_included_packs(struct string_list *include,
 		 * will suffice, so pick the first one.)
 		 */
 		for_each_string_list_item(item, names) {
-			struct generated_pack_data *data = item->util;
-			if (has_pack_ext(data, ".mtimes"))
+			struct generated_pack *pack = item->util;
+			if (generated_pack_has_ext(pack, ".mtimes"))
 				continue;
 
 			strvec_pushf(&cmd.args, "--preferred-pack=pack-%s.pack",
@@ -864,7 +864,7 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 		 */
 		if (local) {
 			item = string_list_append(names, line.buf);
-			item->util = populate_pack_exts(line.buf);
+			item->util = generated_pack_populate(line.buf);
 		}
 	}
 	fclose(out);
@@ -1435,7 +1435,7 @@ int cmd_repack(int argc,
 	 * Ok we have prepared all new packfiles.
 	 */
 	for_each_string_list_item(item, &names) {
-		struct generated_pack_data *data = item->util;
+		struct generated_pack *pack = item->util;
 
 		for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
 			char *fname;
@@ -1443,8 +1443,8 @@ int cmd_repack(int argc,
 			fname = mkpathdup("%s/pack-%s%s",
 					packdir, item->string, exts[ext].name);
 
-			if (data->tempfiles[ext]) {
-				const char *fname_old = get_tempfile_path(data->tempfiles[ext]);
+			if (pack->tempfiles[ext]) {
+				const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
 				struct stat statbuffer;
 
 				if (!stat(fname_old, &statbuffer)) {
@@ -1452,7 +1452,7 @@ int cmd_repack(int argc,
 					chmod(fname_old, statbuffer.st_mode);
 				}
 
-				if (rename_tempfile(&data->tempfiles[ext], fname))
+				if (rename_tempfile(&pack->tempfiles[ext], fname))
 					die_errno(_("renaming pack to '%s' failed"), fname);
 			} else if (!exts[ext].optional)
 				die(_("pack-objects did not write a '%s' file for pack %s-%s"),
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 21/49] builtin/repack.c: factor out "generated_pack_install"

Once all new packs are known to exist, 'repack' installs their contents from their temporary location into their permanent one. This is a semi-involved procedure for each pack, since for each extension (e.g., ".idx", ".pack", ".mtimes", and so on) we have to either:

 - adjust the filemode of the temporary file before renaming it into
   place, or
 - die() if we are missing a non-optional extension, or
 - unlink() any existing file for extensions that we did not generate
   (e.g., if a non-cruft pack we generated was identical to, say, a
   cruft pack which existed at the beginning of the process, we have to
   remove the ".mtimes" file).

Extract this procedure into its own function, and call it "generated_pack_install"(). This will set us up for pulling this function out of the builtin entirely and making it part of the repack.h API, which will be done in a future commit.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 65 ++++++++++++++++++++++++++----------------------
 1 file changed, 35 insertions(+), 30 deletions(-)
Show changes to builtin/repack.c +35 −30
diff --git a/builtin/repack.c b/builtin/repack.c
index a4d80b6b04..bf413a6ee2 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -183,6 +183,38 @@ static int generated_pack_has_ext(const struct generated_pack *pack,
 	BUG("unknown pack extension: '%s'", ext);
 }
 
+static void generated_pack_install(struct generated_pack *pack,
+				   const char *name)
+{
+	int ext;
+	for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
+		char *fname;
+
+		fname = mkpathdup("%s/pack-%s%s", packdir, name,
+				  exts[ext].name);
+
+		if (pack->tempfiles[ext]) {
+			const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
+			struct stat statbuffer;
+
+			if (!stat(fname_old, &statbuffer)) {
+				statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
+				chmod(fname_old, statbuffer.st_mode);
+			}
+
+			if (rename_tempfile(&pack->tempfiles[ext], fname))
+				die_errno(_("renaming pack to '%s' failed"),
+					  fname);
+		} else if (!exts[ext].optional)
+			die(_("pack-objects did not write a '%s' file for pack %s-%s"),
+			    exts[ext].name, packtmp, name);
+		else if (unlink(fname) < 0 && errno != ENOENT)
+			die_errno(_("could not unlink: %s"), fname);
+
+		free(fname);
+	}
+}
+
 static void repack_promisor_objects(struct repository *repo,
 				    const struct pack_objects_args *args,
 				    struct string_list *names)
@@ -1045,7 +1077,7 @@ int cmd_repack(int argc,
 	struct existing_packs existing = EXISTING_PACKS_INIT;
 	struct pack_geometry geometry = { 0 };
 	struct tempfile *refs_snapshot = NULL;
-	int i, ext, ret;
+	int i, ret;
 	int show_progress;
 	char **midx_pack_names = NULL;
 	size_t midx_pack_names_nr = 0;
@@ -1434,35 +1466,8 @@ int cmd_repack(int argc,
 	/*
 	 * Ok we have prepared all new packfiles.
 	 */
-	for_each_string_list_item(item, &names) {
-		struct generated_pack *pack = item->util;
-
-		for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
-			char *fname;
-
-			fname = mkpathdup("%s/pack-%s%s",
-					packdir, item->string, exts[ext].name);
-
-			if (pack->tempfiles[ext]) {
-				const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
-				struct stat statbuffer;
-
-				if (!stat(fname_old, &statbuffer)) {
-					statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
-					chmod(fname_old, statbuffer.st_mode);
-				}
-
-				if (rename_tempfile(&pack->tempfiles[ext], fname))
-					die_errno(_("renaming pack to '%s' failed"), fname);
-			} else if (!exts[ext].optional)
-				die(_("pack-objects did not write a '%s' file for pack %s-%s"),
-				    exts[ext].name, packtmp, item->string);
-			else if (unlink(fname) < 0 && errno != ENOENT)
-				die_errno(_("could not unlink: %s"), fname);
-
-			free(fname);
-		}
-	}
+	for_each_string_list_item(item, &names)
+		generated_pack_install(item->util, item->string);
 	/* End of pack replacement. */
 
 	if (delete_redundant && pack_everything & ALL_INTO_ONE)
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 22/49] builtin/repack.c: pass "packtmp" to `generated_pack_populate()`

In a similar spirit as previous commits, this function needs to know the temporary pack prefix, which it currently accesses through the static "packtmp" variable within builtin/repack.c.

Pass it explicitly as a function parameter to facilitate moving this function out of builtin/repack.c entirely.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
Show changes to builtin/repack.c +4 −3
diff --git a/builtin/repack.c b/builtin/repack.c
index bf413a6ee2..bed902adde 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -150,7 +150,8 @@ struct generated_pack {
 	struct tempfile *tempfiles[ARRAY_SIZE(exts)];
 };
 
-static struct generated_pack *generated_pack_populate(const char *name)
+static struct generated_pack *generated_pack_populate(const char *name,
+						      const char *packtmp)
 {
 	struct stat statbuf;
 	struct strbuf path = STRBUF_INIT;
@@ -271,7 +272,7 @@ static void repack_promisor_objects(struct repository *repo,
 					  line.buf);
 		write_promisor_file(promisor_name, NULL, 0);
 
-		item->util = generated_pack_populate(item->string);
+		item->util = generated_pack_populate(item->string, packtmp);
 
 		free(promisor_name);
 	}
@@ -896,7 +897,7 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 		 */
 		if (local) {
 			item = string_list_append(names, line.buf);
-			item->util = generated_pack_populate(line.buf);
+			item->util = generated_pack_populate(line.buf, packtmp);
 		}
 	}
 	fclose(out);
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 23/49] builtin/repack.c: provide pack locations to `generated_pack_install()`

Repeat what was done in the preceding commit for the `generated_pack_install()` function, which needs both "packdir" and "packtmp".

(As an aside, it is somewhat unfortunate that the final three parameters to this function are all "const char *", making errors like passing "packdir" and "packtmp" in the wrong order easy. We could define a new structure here, but that may be too heavy-handed.)

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +4 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index bed902adde..966db27613 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -185,7 +185,8 @@ static int generated_pack_has_ext(const struct generated_pack *pack,
 }
 
 static void generated_pack_install(struct generated_pack *pack,
-				   const char *name)
+				   const char *name,
+				   const char *packdir, const char *packtmp)
 {
 	int ext;
 	for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
@@ -1468,7 +1469,8 @@ int cmd_repack(int argc,
 	 * Ok we have prepared all new packfiles.
 	 */
 	for_each_string_list_item(item, &names)
-		generated_pack_install(item->util, item->string);
+		generated_pack_install(item->util, item->string, packdir,
+				       packtmp);
 	/* End of pack replacement. */
 
 	if (delete_redundant && pack_everything & ALL_INTO_ONE)
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 24/49] repack: remove 'generated_pack' API from the builtin

Now that we have factored the "generated_pack" API, we can move it to repack.ch, further slimming down builtin/repack.c.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 83 ------------------------------------------------
 repack.c         | 83 ++++++++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  8 +++++
 3 files changed, 91 insertions(+), 83 deletions(-)
Show changes to 3 files +91 −83

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 966db27613..0e11c3b2c9 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -134,89 +134,6 @@ static int write_oid(const struct object_id *oid,
 	return 0;
 }
 
-static struct {
-	const char *name;
-	unsigned optional:1;
-} exts[] = {
-	{".pack"},
-	{".rev", 1},
-	{".mtimes", 1},
-	{".bitmap", 1},
-	{".promisor", 1},
-	{".idx"},
-};
-
-struct generated_pack {
-	struct tempfile *tempfiles[ARRAY_SIZE(exts)];
-};
-
-static struct generated_pack *generated_pack_populate(const char *name,
-						      const char *packtmp)
-{
-	struct stat statbuf;
-	struct strbuf path = STRBUF_INIT;
-	struct generated_pack *pack = xcalloc(1, sizeof(*pack));
-	int i;
-
-	for (i = 0; i < ARRAY_SIZE(exts); i++) {
-		strbuf_reset(&path);
-		strbuf_addf(&path, "%s-%s%s", packtmp, name, exts[i].name);
-
-		if (stat(path.buf, &statbuf))
-			continue;
-
-		pack->tempfiles[i] = register_tempfile(path.buf);
-	}
-
-	strbuf_release(&path);
-	return pack;
-}
-
-static int generated_pack_has_ext(const struct generated_pack *pack,
-				  const char *ext)
-{
-	int i;
-	for (i = 0; i < ARRAY_SIZE(exts); i++) {
-		if (strcmp(exts[i].name, ext))
-			continue;
-		return !!pack->tempfiles[i];
-	}
-	BUG("unknown pack extension: '%s'", ext);
-}
-
-static void generated_pack_install(struct generated_pack *pack,
-				   const char *name,
-				   const char *packdir, const char *packtmp)
-{
-	int ext;
-	for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
-		char *fname;
-
-		fname = mkpathdup("%s/pack-%s%s", packdir, name,
-				  exts[ext].name);
-
-		if (pack->tempfiles[ext]) {
-			const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
-			struct stat statbuffer;
-
-			if (!stat(fname_old, &statbuffer)) {
-				statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
-				chmod(fname_old, statbuffer.st_mode);
-			}
-
-			if (rename_tempfile(&pack->tempfiles[ext], fname))
-				die_errno(_("renaming pack to '%s' failed"),
-					  fname);
-		} else if (!exts[ext].optional)
-			die(_("pack-objects did not write a '%s' file for pack %s-%s"),
-			    exts[ext].name, packtmp, name);
-		else if (unlink(fname) < 0 && errno != ENOENT)
-			die_errno(_("could not unlink: %s"), fname);
-
-		free(fname);
-	}
-}
-
 static void repack_promisor_objects(struct repository *repo,
 				    const struct pack_objects_args *args,
 				    struct string_list *names)
diff --git a/repack.c b/repack.c
index 9182e1c50b..d8afdd352d 100644
--- a/repack.c
+++ b/repack.c
@@ -3,9 +3,11 @@
 #include "midx.h"
 #include "odb.h"
 #include "packfile.h"
+#include "path.h"
 #include "repack.h"
 #include "repository.h"
 #include "run-command.h"
+#include "tempfile.h"
 
 void prepare_pack_objects(struct child_process *cmd,
 			  const struct pack_objects_args *args,
@@ -219,3 +221,84 @@ void existing_packs_release(struct existing_packs *existing)
 	string_list_clear(&existing->non_kept_packs, 0);
 	string_list_clear(&existing->cruft_packs, 0);
 }
+
+static struct {
+	const char *name;
+	unsigned optional:1;
+} exts[] = {
+	{".pack"},
+	{".rev", 1},
+	{".mtimes", 1},
+	{".bitmap", 1},
+	{".promisor", 1},
+	{".idx"},
+};
+
+struct generated_pack {
+	struct tempfile *tempfiles[ARRAY_SIZE(exts)];
+};
+
+struct generated_pack *generated_pack_populate(const char *name,
+					       const char *packtmp)
+{
+	struct stat statbuf;
+	struct strbuf path = STRBUF_INIT;
+	struct generated_pack *pack = xcalloc(1, sizeof(*pack));
+	size_t i;
+
+	for (i = 0; i < ARRAY_SIZE(exts); i++) {
+		strbuf_reset(&path);
+		strbuf_addf(&path, "%s-%s%s", packtmp, name, exts[i].name);
+
+		if (stat(path.buf, &statbuf))
+			continue;
+
+		pack->tempfiles[i] = register_tempfile(path.buf);
+	}
+
+	strbuf_release(&path);
+	return pack;
+}
+
+int generated_pack_has_ext(const struct generated_pack *pack, const char *ext)
+{
+	size_t i;
+	for (i = 0; i < ARRAY_SIZE(exts); i++) {
+		if (strcmp(exts[i].name, ext))
+			continue;
+		return !!pack->tempfiles[i];
+	}
+	BUG("unknown pack extension: '%s'", ext);
+}
+
+void generated_pack_install(struct generated_pack *pack, const char *name,
+			    const char *packdir, const char *packtmp)
+{
+	size_t ext;
+	for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
+		char *fname;
+
+		fname = mkpathdup("%s/pack-%s%s", packdir, name,
+				  exts[ext].name);
+
+		if (pack->tempfiles[ext]) {
+			const char *fname_old = get_tempfile_path(pack->tempfiles[ext]);
+			struct stat statbuffer;
+
+			if (!stat(fname_old, &statbuffer)) {
+				statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
+				chmod(fname_old, statbuffer.st_mode);
+			}
+
+			if (rename_tempfile(&pack->tempfiles[ext], fname))
+				die_errno(_("renaming pack to '%s' failed"),
+					  fname);
+		} else if (!exts[ext].optional)
+			die(_("pack-objects did not write a '%s' file for pack %s-%s"),
+			    exts[ext].name, packtmp, name);
+		else if (unlink(fname) < 0 && errno != ENOENT)
+			die_errno(_("could not unlink: %s"), fname);
+
+		free(fname);
+	}
+}
diff --git a/repack.h b/repack.h
index 19796e2243..f37eb49524 100644
--- a/repack.h
+++ b/repack.h
@@ -66,4 +66,12 @@ void existing_packs_remove_redundant(struct existing_packs *existing,
 				     const char *packdir);
 void existing_packs_release(struct existing_packs *existing);
 
+struct generated_pack;
+
+struct generated_pack *generated_pack_populate(const char *name,
+					       const char *packtmp);
+int generated_pack_has_ext(const struct generated_pack *pack, const char *ext);
+void generated_pack_install(struct generated_pack *pack, const char *name,
+			    const char *packdir, const char *packtmp);
+
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 25/49] builtin/repack.c: pass "packtmp" to `repack_promisor_objects()`

In a similar spirit as previous commit(s), pass the "packtmp" variable to "repack_promisor_objects()" as an explicit parameter of the function, preparing us to move this function in a following commit.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +3 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index 0e11c3b2c9..2c67111b33 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -136,7 +136,8 @@ static int write_oid(const struct object_id *oid,
 
 static void repack_promisor_objects(struct repository *repo,
 				    const struct pack_objects_args *args,
-				    struct string_list *names)
+				    struct string_list *names,
+				    const char *packtmp)
 {
 	struct write_oid_context ctx;
 	struct child_process cmd = CHILD_PROCESS_INIT;
@@ -1199,7 +1200,7 @@ int cmd_repack(int argc,
 		strvec_push(&cmd.args, "--delta-islands");
 
 	if (pack_everything & ALL_INTO_ONE) {
-		repack_promisor_objects(repo, &po_args, &names);
+		repack_promisor_objects(repo, &po_args, &names, packtmp);
 
 		if (existing_packs_has_non_kept(&existing) &&
 		    delete_redundant &&
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 26/49] builtin/repack.c: remove "repack_promisor_objects()" from the builtin

Now that we have properly factored the portion of the builtin which is responsible for repacking promisor objects, we can move that function (and associated dependencies) out of the builtin entirely.

Similar to previous extractions, this function is declared in repack.h, but implemented in a separate repack-promisor.c file. This is done to separate promisor-specific repacking functionality from generic repack utilities (like "existing_packs", and "generated_pack" APIs).

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile          |   1 +
 builtin/repack.c  |  95 ------------------------------------------
 meson.build       |   1 +
 repack-promisor.c | 102 ++++++++++++++++++++++++++++++++++++++++++++++
 repack.h          |   4 ++
 5 files changed, 108 insertions(+), 95 deletions(-)
 create mode 100644 repack-promisor.c
Show changes to 5 files +108 −95

Makefile, builtin/repack.c, meson.build, repack-promisor.c, repack.h

diff --git a/Makefile b/Makefile
index 58fc05030b..01597af359 100644
--- a/Makefile
+++ b/Makefile
@@ -1250,6 +1250,7 @@ LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
+LIB_OBJS += repack-promisor.o
 LIB_OBJS += replace-object.o
 LIB_OBJS += repo-settings.o
 LIB_OBJS += repository.o
diff --git a/builtin/repack.c b/builtin/repack.c
index 2c67111b33..24b5e5049b 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,101 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-struct write_oid_context {
-	struct child_process *cmd;
-	const struct git_hash_algo *algop;
-};
-
-/*
- * Write oid to the given struct child_process's stdin, starting it first if
- * necessary.
- */
-static int write_oid(const struct object_id *oid,
-		     struct packed_git *pack UNUSED,
-		     uint32_t pos UNUSED, void *data)
-{
-	struct write_oid_context *ctx = data;
-	struct child_process *cmd = ctx->cmd;
-
-	if (cmd->in == -1) {
-		if (start_command(cmd))
-			die(_("could not start pack-objects to repack promisor objects"));
-	}
-
-	if (write_in_full(cmd->in, oid_to_hex(oid), ctx->algop->hexsz) < 0 ||
-	    write_in_full(cmd->in, "\n", 1) < 0)
-		die(_("failed to feed promisor objects to pack-objects"));
-	return 0;
-}
-
-static void repack_promisor_objects(struct repository *repo,
-				    const struct pack_objects_args *args,
-				    struct string_list *names,
-				    const char *packtmp)
-{
-	struct write_oid_context ctx;
-	struct child_process cmd = CHILD_PROCESS_INIT;
-	FILE *out;
-	struct strbuf line = STRBUF_INIT;
-
-	prepare_pack_objects(&cmd, args, packtmp);
-	cmd.in = -1;
-
-	/*
-	 * NEEDSWORK: Giving pack-objects only the OIDs without any ordering
-	 * hints may result in suboptimal deltas in the resulting pack. See if
-	 * the OIDs can be sent with fake paths such that pack-objects can use a
-	 * {type -> existing pack order} ordering when computing deltas instead
-	 * of a {type -> size} ordering, which may produce better deltas.
-	 */
-	ctx.cmd = &cmd;
-	ctx.algop = repo->hash_algo;
-	for_each_packed_object(repo, write_oid, &ctx,
-			       FOR_EACH_OBJECT_PROMISOR_ONLY);
-
-	if (cmd.in == -1) {
-		/* No packed objects; cmd was never started */
-		child_process_clear(&cmd);
-		return;
-	}
-
-	close(cmd.in);
-
-	out = xfdopen(cmd.out, "r");
-	while (strbuf_getline_lf(&line, out) != EOF) {
-		struct string_list_item *item;
-		char *promisor_name;
-
-		if (line.len != repo->hash_algo->hexsz)
-			die(_("repack: Expecting full hex object ID lines only from pack-objects."));
-		item = string_list_append(names, line.buf);
-
-		/*
-		 * pack-objects creates the .pack and .idx files, but not the
-		 * .promisor file. Create the .promisor file, which is empty.
-		 *
-		 * NEEDSWORK: fetch-pack sometimes generates non-empty
-		 * .promisor files containing the ref names and associated
-		 * hashes at the point of generation of the corresponding
-		 * packfile, but this would not preserve their contents. Maybe
-		 * concatenate the contents of all .promisor files instead of
-		 * just creating a new empty file.
-		 */
-		promisor_name = mkpathdup("%s-%s.promisor", packtmp,
-					  line.buf);
-		write_promisor_file(promisor_name, NULL, 0);
-
-		item->util = generated_pack_populate(item->string, packtmp);
-
-		free(promisor_name);
-	}
-
-	fclose(out);
-	if (finish_command(&cmd))
-		die(_("could not finish pack-objects to repack promisor objects"));
-	strbuf_release(&line);
-}
-
 struct pack_geometry {
 	struct packed_git **pack;
 	uint32_t pack_nr, pack_alloc;
diff --git a/meson.build b/meson.build
index 4a985ce77e..b37e3d192d 100644
--- a/meson.build
+++ b/meson.build
@@ -463,6 +463,7 @@ libgit_sources = [
   'reftable/writer.c',
   'remote.c',
   'repack.c',
+  'repack-promisor.c',
   'replace-object.c',
   'repo-settings.c',
   'repository.c',
diff --git a/repack-promisor.c b/repack-promisor.c
new file mode 100644
index 0000000000..ee6e0669f6
--- /dev/null
+++ b/repack-promisor.c
@@ -0,0 +1,102 @@
+#include "git-compat-util.h"
+#include "repack.h"
+#include "hex.h"
+#include "pack.h"
+#include "packfile.h"
+#include "path.h"
+#include "repository.h"
+#include "run-command.h"
+
+struct write_oid_context {
+	struct child_process *cmd;
+	const struct git_hash_algo *algop;
+};
+
+/*
+ * Write oid to the given struct child_process's stdin, starting it first if
+ * necessary.
+ */
+static int write_oid(const struct object_id *oid,
+		     struct packed_git *pack UNUSED,
+		     uint32_t pos UNUSED, void *data)
+{
+	struct write_oid_context *ctx = data;
+	struct child_process *cmd = ctx->cmd;
+
+	if (cmd->in == -1) {
+		if (start_command(cmd))
+			die(_("could not start pack-objects to repack promisor objects"));
+	}
+
+	if (write_in_full(cmd->in, oid_to_hex(oid), ctx->algop->hexsz) < 0 ||
+	    write_in_full(cmd->in, "\n", 1) < 0)
+		die(_("failed to feed promisor objects to pack-objects"));
+	return 0;
+}
+
+void repack_promisor_objects(struct repository *repo,
+			     const struct pack_objects_args *args,
+			     struct string_list *names, const char *packtmp)
+{
+	struct write_oid_context ctx;
+	struct child_process cmd = CHILD_PROCESS_INIT;
+	FILE *out;
+	struct strbuf line = STRBUF_INIT;
+
+	prepare_pack_objects(&cmd, args, packtmp);
+	cmd.in = -1;
+
+	/*
+	 * NEEDSWORK: Giving pack-objects only the OIDs without any ordering
+	 * hints may result in suboptimal deltas in the resulting pack. See if
+	 * the OIDs can be sent with fake paths such that pack-objects can use a
+	 * {type -> existing pack order} ordering when computing deltas instead
+	 * of a {type -> size} ordering, which may produce better deltas.
+	 */
+	ctx.cmd = &cmd;
+	ctx.algop = repo->hash_algo;
+	for_each_packed_object(repo, write_oid, &ctx,
+			       FOR_EACH_OBJECT_PROMISOR_ONLY);
+
+	if (cmd.in == -1) {
+		/* No packed objects; cmd was never started */
+		child_process_clear(&cmd);
+		return;
+	}
+
+	close(cmd.in);
+
+	out = xfdopen(cmd.out, "r");
+	while (strbuf_getline_lf(&line, out) != EOF) {
+		struct string_list_item *item;
+		char *promisor_name;
+
+		if (line.len != repo->hash_algo->hexsz)
+			die(_("repack: Expecting full hex object ID lines only from pack-objects."));
+		item = string_list_append(names, line.buf);
+
+		/*
+		 * pack-objects creates the .pack and .idx files, but not the
+		 * .promisor file. Create the .promisor file, which is empty.
+		 *
+		 * NEEDSWORK: fetch-pack sometimes generates non-empty
+		 * .promisor files containing the ref names and associated
+		 * hashes at the point of generation of the corresponding
+		 * packfile, but this would not preserve their contents. Maybe
+		 * concatenate the contents of all .promisor files instead of
+		 * just creating a new empty file.
+		 */
+		promisor_name = mkpathdup("%s-%s.promisor", packtmp,
+					  line.buf);
+		write_promisor_file(promisor_name, NULL, 0);
+
+		item->util = generated_pack_populate(item->string, packtmp);
+
+		free(promisor_name);
+	}
+
+	fclose(out);
+	if (finish_command(&cmd))
+		die(_("could not finish pack-objects to repack promisor objects"));
+	strbuf_release(&line);
+}
diff --git a/repack.h b/repack.h
index f37eb49524..19dc4fd738 100644
--- a/repack.h
+++ b/repack.h
@@ -74,4 +74,8 @@ int generated_pack_has_ext(const struct generated_pack *pack, const char *ext);
 void generated_pack_install(struct generated_pack *pack, const char *name,
 			    const char *packdir, const char *packtmp);
 
+void repack_promisor_objects(struct repository *repo,
+			     const struct pack_objects_args *args,
+			     struct string_list *names, const char *packtmp);
+
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 27/49] builtin/repack.c: rename various pack_geometry functions

Rename functions which work with 'struct pack_geometry' to begin with "pack_geometry_". While we're at it, change `free_pack_geometry()` to instead be named `pack_geometry_release()` to match our conventions, and make clear that that function frees the contents of the struct, not the memory allocated to hold the struct itself.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 52 ++++++++++++++++++++++++------------------------
 1 file changed, 26 insertions(+), 26 deletions(-)
Show changes to builtin/repack.c +26 −26
diff --git a/builtin/repack.c b/builtin/repack.c
index 24b5e5049b..42f05d2ebf 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -115,17 +115,17 @@ struct pack_geometry {
 	int split_factor;
 };
 
-static uint32_t geometry_pack_weight(struct packed_git *p)
+static uint32_t pack_geometry_weight(struct packed_git *p)
 {
 	if (open_pack_index(p))
 		die(_("cannot open index for %s"), p->pack_name);
 	return p->num_objects;
 }
 
-static int geometry_cmp(const void *va, const void *vb)
+static int pack_geometry_cmp(const void *va, const void *vb)
 {
-	uint32_t aw = geometry_pack_weight(*(struct packed_git **)va),
-		 bw = geometry_pack_weight(*(struct packed_git **)vb);
+	uint32_t aw = pack_geometry_weight(*(struct packed_git **)va),
+		 bw = pack_geometry_weight(*(struct packed_git **)vb);
 
 	if (aw < bw)
 		return -1;
@@ -134,7 +134,7 @@ static int geometry_cmp(const void *va, const void *vb)
 	return 0;
 }
 
-static void init_pack_geometry(struct pack_geometry *geometry,
+static void pack_geometry_init(struct pack_geometry *geometry,
 			       struct existing_packs *existing,
 			       const struct pack_objects_args *args)
 {
@@ -184,11 +184,11 @@ static void init_pack_geometry(struct pack_geometry *geometry,
 		geometry->pack_nr++;
 	}
 
-	QSORT(geometry->pack, geometry->pack_nr, geometry_cmp);
+	QSORT(geometry->pack, geometry->pack_nr, pack_geometry_cmp);
 	strbuf_release(&buf);
 }
 
-static void split_pack_geometry(struct pack_geometry *geometry)
+static void pack_geometry_split(struct pack_geometry *geometry)
 {
 	uint32_t i;
 	uint32_t split;
@@ -208,13 +208,13 @@ static void split_pack_geometry(struct pack_geometry *geometry)
 		struct packed_git *prev = geometry->pack[i - 1];
 
 		if (unsigned_mult_overflows(geometry->split_factor,
-					    geometry_pack_weight(prev)))
+					    pack_geometry_weight(prev)))
 			die(_("pack %s too large to consider in geometric "
 			      "progression"),
 			    prev->pack_name);
 
-		if (geometry_pack_weight(ours) <
-		    geometry->split_factor * geometry_pack_weight(prev))
+		if (pack_geometry_weight(ours) <
+		    geometry->split_factor * pack_geometry_weight(prev))
 			break;
 	}
 
@@ -242,9 +242,9 @@ static void split_pack_geometry(struct pack_geometry *geometry)
 	for (i = 0; i < split; i++) {
 		struct packed_git *p = geometry->pack[i];
 
-		if (unsigned_add_overflows(total_size, geometry_pack_weight(p)))
+		if (unsigned_add_overflows(total_size, pack_geometry_weight(p)))
 			die(_("pack %s too large to roll up"), p->pack_name);
-		total_size += geometry_pack_weight(p);
+		total_size += pack_geometry_weight(p);
 	}
 	for (i = split; i < geometry->pack_nr; i++) {
 		struct packed_git *ours = geometry->pack[i];
@@ -253,15 +253,15 @@ static void split_pack_geometry(struct pack_geometry *geometry)
 					    total_size))
 			die(_("pack %s too large to roll up"), ours->pack_name);
 
-		if (geometry_pack_weight(ours) <
+		if (pack_geometry_weight(ours) <
 		    geometry->split_factor * total_size) {
 			if (unsigned_add_overflows(total_size,
-						   geometry_pack_weight(ours)))
+						   pack_geometry_weight(ours)))
 				die(_("pack %s too large to roll up"),
 				    ours->pack_name);
 
 			split++;
-			total_size += geometry_pack_weight(ours);
+			total_size += pack_geometry_weight(ours);
 		} else
 			break;
 	}
@@ -269,7 +269,7 @@ static void split_pack_geometry(struct pack_geometry *geometry)
 	geometry->split = split;
 }
 
-static struct packed_git *get_preferred_pack(struct pack_geometry *geometry)
+static struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry)
 {
 	uint32_t i;
 
@@ -304,9 +304,9 @@ static struct packed_git *get_preferred_pack(struct pack_geometry *geometry)
 	return NULL;
 }
 
-static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
-					    struct string_list *names,
-					    struct existing_packs *existing)
+static void pack_geometry_remove_redundant(struct pack_geometry *geometry,
+					   struct string_list *names,
+					   struct existing_packs *existing)
 {
 	const struct git_hash_algo *algop = existing->repo->hash_algo;
 	struct strbuf buf = STRBUF_INIT;
@@ -332,7 +332,7 @@ static void geometry_remove_redundant_packs(struct pack_geometry *geometry,
 	strbuf_release(&buf);
 }
 
-static void free_pack_geometry(struct pack_geometry *geometry)
+static void pack_geometry_release(struct pack_geometry *geometry)
 {
 	if (!geometry)
 		return;
@@ -599,7 +599,7 @@ static int write_midx_included_packs(struct string_list *include,
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
-	struct packed_git *preferred = get_preferred_pack(geometry);
+	struct packed_git *preferred = pack_geometry_preferred_pack(geometry);
 	FILE *in;
 	int ret;
 
@@ -1063,8 +1063,8 @@ int cmd_repack(int argc,
 	if (geometry.split_factor) {
 		if (pack_everything)
 			die(_("options '%s' and '%s' cannot be used together"), "--geometric", "-A/-a");
-		init_pack_geometry(&geometry, &existing, &po_args);
-		split_pack_geometry(&geometry);
+		pack_geometry_init(&geometry, &existing, &po_args);
+		pack_geometry_split(&geometry);
 	}
 
 	prepare_pack_objects(&cmd, &po_args, packtmp);
@@ -1324,8 +1324,8 @@ int cmd_repack(int argc,
 		existing_packs_remove_redundant(&existing, packdir);
 
 		if (geometry.split_factor)
-			geometry_remove_redundant_packs(&geometry, &names,
-							&existing);
+			pack_geometry_remove_redundant(&geometry, &names,
+						       &existing);
 		if (show_progress)
 			opts |= PRUNE_PACKED_VERBOSE;
 		prune_packed_objects(opts);
@@ -1352,7 +1352,7 @@ int cmd_repack(int argc,
 	string_list_clear(&keep_pack_list, 0);
 	string_list_clear(&names, 1);
 	existing_packs_release(&existing);
-	free_pack_geometry(&geometry);
+	pack_geometry_release(&geometry);
 	for (size_t i = 0; i < midx_pack_names_nr; i++)
 		free(midx_pack_names[i]);
 	free(midx_pack_names);
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 28/49] builtin/repack.c: pass 'pack_kept_objects' to `pack_geometry_init()`

Prepare to move pack_geometry-related APIs to their own compilation unit by passing in the static "pack_kept_objects" variable directly as a parameter to the 'pack_geometry_init()' function.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +4 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index 42f05d2ebf..ac8c80d0a5 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -136,7 +136,8 @@ static int pack_geometry_cmp(const void *va, const void *vb)
 
 static void pack_geometry_init(struct pack_geometry *geometry,
 			       struct existing_packs *existing,
-			       const struct pack_objects_args *args)
+			       const struct pack_objects_args *args,
+			       int pack_kept_objects)
 {
 	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
@@ -1063,7 +1064,8 @@ int cmd_repack(int argc,
 	if (geometry.split_factor) {
 		if (pack_everything)
 			die(_("options '%s' and '%s' cannot be used together"), "--geometric", "-A/-a");
-		pack_geometry_init(&geometry, &existing, &po_args);
+		pack_geometry_init(&geometry, &existing, &po_args,
+				   pack_kept_objects);
 		pack_geometry_split(&geometry);
 	}
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 29/49] builtin/repack.c: pass 'packdir' to `pack_geometry_remove_redundant()`

For similar reasons as the preceding commit, pass the "packdir" variable directly to `pack_geometry_remove_redundant()` as a parameter to the function.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
Show changes to builtin/repack.c +3 −2
diff --git a/builtin/repack.c b/builtin/repack.c
index ac8c80d0a5..f6d04b33a7 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -307,7 +307,8 @@ static struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geo
 
 static void pack_geometry_remove_redundant(struct pack_geometry *geometry,
 					   struct string_list *names,
-					   struct existing_packs *existing)
+					   struct existing_packs *existing,
+					   const char *packdir)
 {
 	const struct git_hash_algo *algop = existing->repo->hash_algo;
 	struct strbuf buf = STRBUF_INIT;
@@ -1327,7 +1328,7 @@ int cmd_repack(int argc,
 
 		if (geometry.split_factor)
 			pack_geometry_remove_redundant(&geometry, &names,
-						       &existing);
+						       &existing, packdir);
 		if (show_progress)
 			opts |= PRUNE_PACKED_VERBOSE;
 		prune_packed_objects(opts);
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 30/49] repack: remove pack_geometry API from the builtin

Now that the pack_geometry API is fully factored and isolated from the rest of the builtin, declare it within repack.h and move its implementation to "repack-geometry.c" as a separate component.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile          |   1 +
 builtin/repack.c  | 235 ----------------------------------------------
 meson.build       |   1 +
 repack-geometry.c | 234 +++++++++++++++++++++++++++++++++++++++++++++
 repack.h          |  20 ++++
 5 files changed, 256 insertions(+), 235 deletions(-)
 create mode 100644 repack-geometry.c
Show changes to 5 files +256 −235

Makefile, builtin/repack.c, meson.build, repack-geometry.c, repack.h

diff --git a/Makefile b/Makefile
index 01597af359..a1ccac235d 100644
--- a/Makefile
+++ b/Makefile
@@ -1250,6 +1250,7 @@ LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
+LIB_OBJS += repack-geometry.o
 LIB_OBJS += repack-promisor.o
 LIB_OBJS += replace-object.o
 LIB_OBJS += repo-settings.o
diff --git a/builtin/repack.c b/builtin/repack.c
index f6d04b33a7..e2313c80c3 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,241 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-struct pack_geometry {
-	struct packed_git **pack;
-	uint32_t pack_nr, pack_alloc;
-	uint32_t split;
-
-	int split_factor;
-};
-
-static uint32_t pack_geometry_weight(struct packed_git *p)
-{
-	if (open_pack_index(p))
-		die(_("cannot open index for %s"), p->pack_name);
-	return p->num_objects;
-}
-
-static int pack_geometry_cmp(const void *va, const void *vb)
-{
-	uint32_t aw = pack_geometry_weight(*(struct packed_git **)va),
-		 bw = pack_geometry_weight(*(struct packed_git **)vb);
-
-	if (aw < bw)
-		return -1;
-	if (aw > bw)
-		return 1;
-	return 0;
-}
-
-static void pack_geometry_init(struct pack_geometry *geometry,
-			       struct existing_packs *existing,
-			       const struct pack_objects_args *args,
-			       int pack_kept_objects)
-{
-	struct packfile_store *packs = existing->repo->objects->packfiles;
-	struct packed_git *p;
-	struct strbuf buf = STRBUF_INIT;
-
-	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
-		if (args->local && !p->pack_local)
-			/*
-			 * When asked to only repack local packfiles we skip
-			 * over any packfiles that are borrowed from alternate
-			 * object directories.
-			 */
-			continue;
-
-		if (!pack_kept_objects) {
-			/*
-			 * Any pack that has its pack_keep bit set will
-			 * appear in existing->kept_packs below, but
-			 * this saves us from doing a more expensive
-			 * check.
-			 */
-			if (p->pack_keep)
-				continue;
-
-			/*
-			 * The pack may be kept via the --keep-pack
-			 * option; check 'existing->kept_packs' to
-			 * determine whether to ignore it.
-			 */
-			strbuf_reset(&buf);
-			strbuf_addstr(&buf, pack_basename(p));
-			strbuf_strip_suffix(&buf, ".pack");
-
-			if (string_list_has_string(&existing->kept_packs, buf.buf))
-				continue;
-		}
-		if (p->is_cruft)
-			continue;
-
-		ALLOC_GROW(geometry->pack,
-			   geometry->pack_nr + 1,
-			   geometry->pack_alloc);
-
-		geometry->pack[geometry->pack_nr] = p;
-		geometry->pack_nr++;
-	}
-
-	QSORT(geometry->pack, geometry->pack_nr, pack_geometry_cmp);
-	strbuf_release(&buf);
-}
-
-static void pack_geometry_split(struct pack_geometry *geometry)
-{
-	uint32_t i;
-	uint32_t split;
-	off_t total_size = 0;
-
-	if (!geometry->pack_nr) {
-		geometry->split = geometry->pack_nr;
-		return;
-	}
-
-	/*
-	 * First, count the number of packs (in descending order of size) which
-	 * already form a geometric progression.
-	 */
-	for (i = geometry->pack_nr - 1; i > 0; i--) {
-		struct packed_git *ours = geometry->pack[i];
-		struct packed_git *prev = geometry->pack[i - 1];
-
-		if (unsigned_mult_overflows(geometry->split_factor,
-					    pack_geometry_weight(prev)))
-			die(_("pack %s too large to consider in geometric "
-			      "progression"),
-			    prev->pack_name);
-
-		if (pack_geometry_weight(ours) <
-		    geometry->split_factor * pack_geometry_weight(prev))
-			break;
-	}
-
-	split = i;
-
-	if (split) {
-		/*
-		 * Move the split one to the right, since the top element in the
-		 * last-compared pair can't be in the progression. Only do this
-		 * when we split in the middle of the array (otherwise if we got
-		 * to the end, then the split is in the right place).
-		 */
-		split++;
-	}
-
-	/*
-	 * Then, anything to the left of 'split' must be in a new pack. But,
-	 * creating that new pack may cause packs in the heavy half to no longer
-	 * form a geometric progression.
-	 *
-	 * Compute an expected size of the new pack, and then determine how many
-	 * packs in the heavy half need to be joined into it (if any) to restore
-	 * the geometric progression.
-	 */
-	for (i = 0; i < split; i++) {
-		struct packed_git *p = geometry->pack[i];
-
-		if (unsigned_add_overflows(total_size, pack_geometry_weight(p)))
-			die(_("pack %s too large to roll up"), p->pack_name);
-		total_size += pack_geometry_weight(p);
-	}
-	for (i = split; i < geometry->pack_nr; i++) {
-		struct packed_git *ours = geometry->pack[i];
-
-		if (unsigned_mult_overflows(geometry->split_factor,
-					    total_size))
-			die(_("pack %s too large to roll up"), ours->pack_name);
-
-		if (pack_geometry_weight(ours) <
-		    geometry->split_factor * total_size) {
-			if (unsigned_add_overflows(total_size,
-						   pack_geometry_weight(ours)))
-				die(_("pack %s too large to roll up"),
-				    ours->pack_name);
-
-			split++;
-			total_size += pack_geometry_weight(ours);
-		} else
-			break;
-	}
-
-	geometry->split = split;
-}
-
-static struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry)
-{
-	uint32_t i;
-
-	if (!geometry) {
-		/*
-		 * No geometry means either an all-into-one repack (in which
-		 * case there is only one pack left and it is the largest) or an
-		 * incremental one.
-		 *
-		 * If repacking incrementally, then we could check the size of
-		 * all packs to determine which should be preferred, but leave
-		 * this for later.
-		 */
-		return NULL;
-	}
-	if (geometry->split == geometry->pack_nr)
-		return NULL;
-
-	/*
-	 * The preferred pack is the largest pack above the split line. In
-	 * other words, it is the largest pack that does not get rolled up in
-	 * the geometric repack.
-	 */
-	for (i = geometry->pack_nr; i > geometry->split; i--)
-		/*
-		 * A pack that is not local would never be included in a
-		 * multi-pack index. We thus skip over any non-local packs.
-		 */
-		if (geometry->pack[i - 1]->pack_local)
-			return geometry->pack[i - 1];
-
-	return NULL;
-}
-
-static void pack_geometry_remove_redundant(struct pack_geometry *geometry,
-					   struct string_list *names,
-					   struct existing_packs *existing,
-					   const char *packdir)
-{
-	const struct git_hash_algo *algop = existing->repo->hash_algo;
-	struct strbuf buf = STRBUF_INIT;
-	uint32_t i;
-
-	for (i = 0; i < geometry->split; i++) {
-		struct packed_git *p = geometry->pack[i];
-		if (string_list_has_string(names, hash_to_hex_algop(p->hash,
-								    algop)))
-			continue;
-
-		strbuf_reset(&buf);
-		strbuf_addstr(&buf, pack_basename(p));
-		strbuf_strip_suffix(&buf, ".pack");
-
-		if ((p->pack_keep) ||
-		    (string_list_has_string(&existing->kept_packs, buf.buf)))
-			continue;
-
-		repack_remove_redundant_pack(existing->repo, packdir, buf.buf);
-	}
-
-	strbuf_release(&buf);
-}
-
-static void pack_geometry_release(struct pack_geometry *geometry)
-{
-	if (!geometry)
-		return;
-
-	free(geometry->pack);
-}
-
 static int midx_has_unknown_packs(char **midx_pack_names,
 				  size_t midx_pack_names_nr,
 				  struct string_list *include,
diff --git a/meson.build b/meson.build
index b37e3d192d..9afd45ea45 100644
--- a/meson.build
+++ b/meson.build
@@ -463,6 +463,7 @@ libgit_sources = [
   'reftable/writer.c',
   'remote.c',
   'repack.c',
+  'repack-geometry.c',
   'repack-promisor.c',
   'replace-object.c',
   'repo-settings.c',
diff --git a/repack-geometry.c b/repack-geometry.c
new file mode 100644
index 0000000000..f58f1fc7f0
--- /dev/null
+++ b/repack-geometry.c
@@ -0,0 +1,234 @@
+#define DISABLE_SIGN_COMPARE_WARNINGS
+
+#include "git-compat-util.h"
+#include "repack.h"
+#include "repository.h"
+#include "hex.h"
+#include "packfile.h"
+
+static uint32_t pack_geometry_weight(struct packed_git *p)
+{
+	if (open_pack_index(p))
+		die(_("cannot open index for %s"), p->pack_name);
+	return p->num_objects;
+}
+
+static int pack_geometry_cmp(const void *va, const void *vb)
+{
+	uint32_t aw = pack_geometry_weight(*(struct packed_git **)va),
+		 bw = pack_geometry_weight(*(struct packed_git **)vb);
+
+	if (aw < bw)
+		return -1;
+	if (aw > bw)
+		return 1;
+	return 0;
+}
+
+void pack_geometry_init(struct pack_geometry *geometry,
+			struct existing_packs *existing,
+			const struct pack_objects_args *args,
+			int pack_kept_objects)
+{
+	struct packfile_store *packs = existing->repo->objects->packfiles;
+	struct packed_git *p;
+	struct strbuf buf = STRBUF_INIT;
+
+	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
+		if (args->local && !p->pack_local)
+			/*
+			 * When asked to only repack local packfiles we skip
+			 * over any packfiles that are borrowed from alternate
+			 * object directories.
+			 */
+			continue;
+
+		if (!pack_kept_objects) {
+			/*
+			 * Any pack that has its pack_keep bit set will
+			 * appear in existing->kept_packs below, but
+			 * this saves us from doing a more expensive
+			 * check.
+			 */
+			if (p->pack_keep)
+				continue;
+
+			/*
+			 * The pack may be kept via the --keep-pack
+			 * option; check 'existing->kept_packs' to
+			 * determine whether to ignore it.
+			 */
+			strbuf_reset(&buf);
+			strbuf_addstr(&buf, pack_basename(p));
+			strbuf_strip_suffix(&buf, ".pack");
+
+			if (string_list_has_string(&existing->kept_packs, buf.buf))
+				continue;
+		}
+		if (p->is_cruft)
+			continue;
+
+		ALLOC_GROW(geometry->pack,
+			   geometry->pack_nr + 1,
+			   geometry->pack_alloc);
+
+		geometry->pack[geometry->pack_nr] = p;
+		geometry->pack_nr++;
+	}
+
+	QSORT(geometry->pack, geometry->pack_nr, pack_geometry_cmp);
+	strbuf_release(&buf);
+}
+
+void pack_geometry_split(struct pack_geometry *geometry)
+{
+	uint32_t i;
+	uint32_t split;
+	off_t total_size = 0;
+
+	if (!geometry->pack_nr) {
+		geometry->split = geometry->pack_nr;
+		return;
+	}
+
+	/*
+	 * First, count the number of packs (in descending order of size) which
+	 * already form a geometric progression.
+	 */
+	for (i = geometry->pack_nr - 1; i > 0; i--) {
+		struct packed_git *ours = geometry->pack[i];
+		struct packed_git *prev = geometry->pack[i - 1];
+
+		if (unsigned_mult_overflows(geometry->split_factor,
+					    pack_geometry_weight(prev)))
+			die(_("pack %s too large to consider in geometric "
+			      "progression"),
+			    prev->pack_name);
+
+		if (pack_geometry_weight(ours) <
+		    geometry->split_factor * pack_geometry_weight(prev))
+			break;
+	}
+
+	split = i;
+
+	if (split) {
+		/*
+		 * Move the split one to the right, since the top element in the
+		 * last-compared pair can't be in the progression. Only do this
+		 * when we split in the middle of the array (otherwise if we got
+		 * to the end, then the split is in the right place).
+		 */
+		split++;
+	}
+
+	/*
+	 * Then, anything to the left of 'split' must be in a new pack. But,
+	 * creating that new pack may cause packs in the heavy half to no longer
+	 * form a geometric progression.
+	 *
+	 * Compute an expected size of the new pack, and then determine how many
+	 * packs in the heavy half need to be joined into it (if any) to restore
+	 * the geometric progression.
+	 */
+	for (i = 0; i < split; i++) {
+		struct packed_git *p = geometry->pack[i];
+
+		if (unsigned_add_overflows(total_size, pack_geometry_weight(p)))
+			die(_("pack %s too large to roll up"), p->pack_name);
+		total_size += pack_geometry_weight(p);
+	}
+	for (i = split; i < geometry->pack_nr; i++) {
+		struct packed_git *ours = geometry->pack[i];
+
+		if (unsigned_mult_overflows(geometry->split_factor,
+					    total_size))
+			die(_("pack %s too large to roll up"), ours->pack_name);
+
+		if (pack_geometry_weight(ours) <
+		    geometry->split_factor * total_size) {
+			if (unsigned_add_overflows(total_size,
+						   pack_geometry_weight(ours)))
+				die(_("pack %s too large to roll up"),
+				    ours->pack_name);
+
+			split++;
+			total_size += pack_geometry_weight(ours);
+		} else
+			break;
+	}
+
+	geometry->split = split;
+}
+
+struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry)
+{
+	uint32_t i;
+
+	if (!geometry) {
+		/*
+		 * No geometry means either an all-into-one repack (in which
+		 * case there is only one pack left and it is the largest) or an
+		 * incremental one.
+		 *
+		 * If repacking incrementally, then we could check the size of
+		 * all packs to determine which should be preferred, but leave
+		 * this for later.
+		 */
+		return NULL;
+	}
+	if (geometry->split == geometry->pack_nr)
+		return NULL;
+
+	/*
+	 * The preferred pack is the largest pack above the split line. In
+	 * other words, it is the largest pack that does not get rolled up in
+	 * the geometric repack.
+	 */
+	for (i = geometry->pack_nr; i > geometry->split; i--)
+		/*
+		 * A pack that is not local would never be included in a
+		 * multi-pack index. We thus skip over any non-local packs.
+		 */
+		if (geometry->pack[i - 1]->pack_local)
+			return geometry->pack[i - 1];
+
+	return NULL;
+}
+
+void pack_geometry_remove_redundant(struct pack_geometry *geometry,
+				    struct string_list *names,
+				    struct existing_packs *existing,
+				    const char *packdir)
+{
+	const struct git_hash_algo *algop = existing->repo->hash_algo;
+	struct strbuf buf = STRBUF_INIT;
+	uint32_t i;
+
+	for (i = 0; i < geometry->split; i++) {
+		struct packed_git *p = geometry->pack[i];
+		if (string_list_has_string(names, hash_to_hex_algop(p->hash,
+								    algop)))
+			continue;
+
+		strbuf_reset(&buf);
+		strbuf_addstr(&buf, pack_basename(p));
+		strbuf_strip_suffix(&buf, ".pack");
+
+		if ((p->pack_keep) ||
+		    (string_list_has_string(&existing->kept_packs, buf.buf)))
+			continue;
+
+		repack_remove_redundant_pack(existing->repo, packdir, buf.buf);
+	}
+
+	strbuf_release(&buf);
+}
+
+void pack_geometry_release(struct pack_geometry *geometry)
+{
+	if (!geometry)
+		return;
+
+	free(geometry->pack);
+}
diff --git a/repack.h b/repack.h
index 19dc4fd738..cea7969ae4 100644
--- a/repack.h
+++ b/repack.h
@@ -78,4 +78,24 @@ void repack_promisor_objects(struct repository *repo,
 			     const struct pack_objects_args *args,
 			     struct string_list *names, const char *packtmp);
 
+struct pack_geometry {
+	struct packed_git **pack;
+	uint32_t pack_nr, pack_alloc;
+	uint32_t split;
+
+	int split_factor;
+};
+
+void pack_geometry_init(struct pack_geometry *geometry,
+			struct existing_packs *existing,
+			const struct pack_objects_args *args,
+			int pack_kept_objects);
+void pack_geometry_split(struct pack_geometry *geometry);
+struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry);
+void pack_geometry_remove_redundant(struct pack_geometry *geometry,
+				    struct string_list *names,
+				    struct existing_packs *existing,
+				    const char *packdir);
+void pack_geometry_release(struct pack_geometry *geometry);
+
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 31/49] builtin/repack.c: remove ref snapshotting from builtin

When writing a MIDX, 'git repack' takes a snapshot of the repository's references and writes the result out to a file, which it then passes to 'git multi-pack-index write' via the '--refs-snapshot'.

This is done in order to make bitmap selections with respect to what we are packing, thus avoiding a race where an incoming reference update causes us to try and write a bitmap for a commit not present in the MIDX.

Extract this functionality out into a new repack-midx.c compilation unit, and expose the necessary functions via the repack.h API.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile         |  1 +
 builtin/repack.c | 68 ------------------------------------------
 meson.build      |  1 +
 repack-midx.c    | 77 ++++++++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  4 +++
 5 files changed, 83 insertions(+), 68 deletions(-)
 create mode 100644 repack-midx.c
Show changes to 5 files +83 −68

Makefile, builtin/repack.c, meson.build, repack-midx.c, repack.h

diff --git a/Makefile b/Makefile
index a1ccac235d..4b0a122428 100644
--- a/Makefile
+++ b/Makefile
@@ -1251,6 +1251,7 @@ LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
 LIB_OBJS += repack-geometry.o
+LIB_OBJS += repack-midx.o
 LIB_OBJS += repack-promisor.o
 LIB_OBJS += replace-object.o
 LIB_OBJS += repo-settings.o
diff --git a/builtin/repack.c b/builtin/repack.c
index e2313c80c3..7713721826 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -178,74 +178,6 @@ static int midx_has_unknown_packs(char **midx_pack_names,
 	return 0;
 }
 
-struct midx_snapshot_ref_data {
-	struct repository *repo;
-	struct tempfile *f;
-	struct oidset seen;
-	int preferred;
-};
-
-static int midx_snapshot_ref_one(const char *refname UNUSED,
-				 const char *referent UNUSED,
-				 const struct object_id *oid,
-				 int flag UNUSED, void *_data)
-{
-	struct midx_snapshot_ref_data *data = _data;
-	struct object_id peeled;
-
-	if (!peel_iterated_oid(data->repo, oid, &peeled))
-		oid = &peeled;
-
-	if (oidset_insert(&data->seen, oid))
-		return 0; /* already seen */
-
-	if (odb_read_object_info(data->repo->objects, oid, NULL) != OBJ_COMMIT)
-		return 0;
-
-	fprintf(data->f->fp, "%s%s\n", data->preferred ? "+" : "",
-		oid_to_hex(oid));
-
-	return 0;
-}
-
-static void midx_snapshot_refs(struct repository *repo, struct tempfile *f)
-{
-	struct midx_snapshot_ref_data data;
-	const struct string_list *preferred = bitmap_preferred_tips(repo);
-
-	data.repo = repo;
-	data.f = f;
-	data.preferred = 0;
-	oidset_init(&data.seen, 0);
-
-	if (!fdopen_tempfile(f, "w"))
-		 die(_("could not open tempfile %s for writing"),
-		     get_tempfile_path(f));
-
-	if (preferred) {
-		struct string_list_item *item;
-
-		data.preferred = 1;
-		for_each_string_list_item(item, preferred)
-			refs_for_each_ref_in(get_main_ref_store(repo),
-					     item->string,
-					     midx_snapshot_ref_one, &data);
-		data.preferred = 0;
-	}
-
-	refs_for_each_ref(get_main_ref_store(repo),
-			  midx_snapshot_ref_one, &data);
-
-	if (close_tempfile_gently(f)) {
-		int save_errno = errno;
-		delete_tempfile(&f);
-		errno = save_errno;
-		die_errno(_("could not close refs snapshot tempfile"));
-	}
-
-	oidset_clear(&data.seen);
-}
-
 static void midx_included_packs(struct string_list *include,
 				struct existing_packs *existing,
 				char **midx_pack_names,
diff --git a/meson.build b/meson.build
index 9afd45ea45..0373d5c454 100644
--- a/meson.build
+++ b/meson.build
@@ -464,6 +464,7 @@ libgit_sources = [
   'remote.c',
   'repack.c',
   'repack-geometry.c',
+  'repack-midx.c',
   'repack-promisor.c',
   'replace-object.c',
   'repo-settings.c',
diff --git a/repack-midx.c b/repack-midx.c
new file mode 100644
index 0000000000..354df729a5
--- /dev/null
+++ b/repack-midx.c
@@ -0,0 +1,77 @@
+#include "git-compat-util.h"
+#include "repack.h"
+#include "hash.h"
+#include "hex.h"
+#include "odb.h"
+#include "oidset.h"
+#include "pack-bitmap.h"
+#include "refs.h"
+#include "tempfile.h"
+
+struct midx_snapshot_ref_data {
+	struct repository *repo;
+	struct tempfile *f;
+	struct oidset seen;
+	int preferred;
+};
+
+static int midx_snapshot_ref_one(const char *refname UNUSED,
+				 const char *referent UNUSED,
+				 const struct object_id *oid,
+				 int flag UNUSED, void *_data)
+{
+	struct midx_snapshot_ref_data *data = _data;
+	struct object_id peeled;
+
+	if (!peel_iterated_oid(data->repo, oid, &peeled))
+		oid = &peeled;
+
+	if (oidset_insert(&data->seen, oid))
+		return 0; /* already seen */
+
+	if (odb_read_object_info(data->repo->objects, oid, NULL) != OBJ_COMMIT)
+		return 0;
+
+	fprintf(data->f->fp, "%s%s\n", data->preferred ? "+" : "",
+		oid_to_hex(oid));
+
+	return 0;
+}
+
+void midx_snapshot_refs(struct repository *repo, struct tempfile *f)
+{
+	struct midx_snapshot_ref_data data;
+	const struct string_list *preferred = bitmap_preferred_tips(repo);
+
+	data.repo = repo;
+	data.f = f;
+	data.preferred = 0;
+	oidset_init(&data.seen, 0);
+
+	if (!fdopen_tempfile(f, "w"))
+		 die(_("could not open tempfile %s for writing"),
+		     get_tempfile_path(f));
+
+	if (preferred) {
+		struct string_list_item *item;
+
+		data.preferred = 1;
+		for_each_string_list_item(item, preferred)
+			refs_for_each_ref_in(get_main_ref_store(repo),
+					     item->string,
+					     midx_snapshot_ref_one, &data);
+		data.preferred = 0;
+	}
+
+	refs_for_each_ref(get_main_ref_store(repo),
+			  midx_snapshot_ref_one, &data);
+
+	if (close_tempfile_gently(f)) {
+		int save_errno = errno;
+		delete_tempfile(&f);
+		errno = save_errno;
+		die_errno(_("could not close refs snapshot tempfile"));
+	}
+
+	oidset_clear(&data.seen);
+}
diff --git a/repack.h b/repack.h
index cea7969ae4..803e129224 100644
--- a/repack.h
+++ b/repack.h
@@ -98,4 +98,8 @@ void pack_geometry_remove_redundant(struct pack_geometry *geometry,
 				    const char *packdir);
 void pack_geometry_release(struct pack_geometry *geometry);
 
+struct tempfile;
+
+void midx_snapshot_refs(struct repository *repo, struct tempfile *f);
+
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 32/49] builtin/repack.c: extract opts struct for 'write_midx_included_packs()'

The function 'write_midx_included_packs()', which is responsible for writing a new MIDX with a given set of included packs, currently takes a list of six arguments.

In order to extract this function out of the builtin, we have to pass in a few additional parameters, like 'midx_must_contain_cruft' and 'packdir', which are currently declared as static variables within the builtin/repack.c compilation unit.

Instead of adding additional parameters to `write_midx_included_packs()` extract out an "opts" struct that names these parameters, and pass a pointer to that, making it less cumbersome to add additional parameters.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 52 +++++++++++++++++++++++++++++++-----------------
 1 file changed, 34 insertions(+), 18 deletions(-)
Show changes to builtin/repack.c +34 −18
diff --git a/builtin/repack.c b/builtin/repack.c
index 7713721826..1a79d1d834 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,6 +107,17 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
+struct repack_write_midx_opts {
+	struct string_list *include;
+	struct pack_geometry *geometry;
+	struct string_list *names;
+	const char *refs_snapshot;
+	const char *packdir;
+	int show_progress;
+	int write_bitmaps;
+	int midx_must_contain_cruft;
+};
+
 static int midx_has_unknown_packs(char **midx_pack_names,
 				  size_t midx_pack_names_nr,
 				  struct string_list *include,
@@ -290,19 +301,15 @@ static void midx_included_packs(struct string_list *include,
 	strbuf_release(&buf);
 }
 
-static int write_midx_included_packs(struct string_list *include,
-				     struct pack_geometry *geometry,
-				     struct string_list *names,
-				     const char *refs_snapshot,
-				     int show_progress, int write_bitmaps)
+static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
-	struct packed_git *preferred = pack_geometry_preferred_pack(geometry);
+	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
 	FILE *in;
 	int ret;
 
-	if (!include->nr)
+	if (!opts->include->nr)
 		return 0;
 
 	cmd.in = -1;
@@ -311,18 +318,18 @@ static int write_midx_included_packs(struct string_list *include,
 	strvec_push(&cmd.args, "multi-pack-index");
 	strvec_pushl(&cmd.args, "write", "--stdin-packs", NULL);
 
-	if (show_progress)
+	if (opts->show_progress)
 		strvec_push(&cmd.args, "--progress");
 	else
 		strvec_push(&cmd.args, "--no-progress");
 
-	if (write_bitmaps)
+	if (opts->write_bitmaps)
 		strvec_push(&cmd.args, "--bitmap");
 
 	if (preferred)
 		strvec_pushf(&cmd.args, "--preferred-pack=%s",
 			     pack_basename(preferred));
-	else if (names->nr) {
+	else if (opts->names->nr) {
 		/* The largest pack was repacked, meaning that either
 		 * one or two packs exist depending on whether the
 		 * repository has a cruft pack or not.
@@ -335,7 +342,7 @@ static int write_midx_included_packs(struct string_list *include,
 		 * `--max-pack-size` was given, but any one of them
 		 * will suffice, so pick the first one.)
 		 */
-		for_each_string_list_item(item, names) {
+		for_each_string_list_item(item, opts->names) {
 			struct generated_pack *pack = item->util;
 			if (generated_pack_has_ext(pack, ".mtimes"))
 				continue;
@@ -355,15 +362,16 @@ static int write_midx_included_packs(struct string_list *include,
 		;
 	}
 
-	if (refs_snapshot)
-		strvec_pushf(&cmd.args, "--refs-snapshot=%s", refs_snapshot);
+	if (opts->refs_snapshot)
+		strvec_pushf(&cmd.args, "--refs-snapshot=%s",
+			     opts->refs_snapshot);
 
 	ret = start_command(&cmd);
 	if (ret)
 		return ret;
 
 	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, include)
+	for_each_string_list_item(item, opts->include)
 		fprintf(in, "%s\n", item->string);
 	fclose(in);
 
@@ -1001,15 +1009,23 @@ int cmd_repack(int argc,
 
 	if (write_midx) {
 		struct string_list include = STRING_LIST_INIT_DUP;
+		struct repack_write_midx_opts opts = {
+			.include = &include,
+			.geometry = &geometry,
+			.names = &names,
+			.refs_snapshot = refs_snapshot ? get_tempfile_path(refs_snapshot) : NULL,
+			.packdir = packdir,
+			.show_progress = show_progress,
+			.write_bitmaps = write_bitmaps > 0,
+			.midx_must_contain_cruft = midx_must_contain_cruft
+		};
 		midx_included_packs(&include, &existing, midx_pack_names,
 				    midx_pack_names_nr, &names, &geometry);
 
-		ret = write_midx_included_packs(&include, &geometry, &names,
-						refs_snapshot ? get_tempfile_path(refs_snapshot) : NULL,
-						show_progress, write_bitmaps > 0);
+		ret = write_midx_included_packs(&opts);
 
 		if (!ret && write_bitmaps)
-			remove_redundant_bitmaps(&include, packdir);
+			remove_redundant_bitmaps(&include, opts.packdir);
 
 		string_list_clear(&include, 0);
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 33/49] builtin/repack.c: use a string_list for 'midx_pack_names'

When writing a new MIDX, repack must determine whether or not there are any packs in the MIDX it is replacing (if one exists) that are not somehow represented in the new MIDX (e.g., either by preserving the pack verbatim, or rolling it up as part of a geometric repack, etc.).

In order to do this, it keeps track of a list of pack names from the MIDX present in the repository at the start of the repack operation. Since we manipulate and close the object store, we cannot rely on the repository's in-core representation of the MIDX, since this is subject to change and/or go away.

When this behavior was introduced in 5ee86c273b (repack: exclude cruft pack(s) from the MIDX where possible, 2025-06-23), we maintained an array of character pointers instead of using a convenience API, such as string-list.h.

Store the list of MIDX pack names in a string_list, thereby reducing the number of parameters we have to pass to `midx_has_unknown_packs()`.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 40 +++++++++++++++++-----------------------
 1 file changed, 17 insertions(+), 23 deletions(-)
Show changes to builtin/repack.c +17 −23
diff --git a/builtin/repack.c b/builtin/repack.c
index 1a79d1d834..dda533f171 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -118,18 +118,17 @@ struct repack_write_midx_opts {
 	int midx_must_contain_cruft;
 };
 
-static int midx_has_unknown_packs(char **midx_pack_names,
-				  size_t midx_pack_names_nr,
+static int midx_has_unknown_packs(struct string_list *midx_pack_names,
 				  struct string_list *include,
 				  struct pack_geometry *geometry,
 				  struct existing_packs *existing)
 {
-	size_t i;
+	struct string_list_item *item;
 
 	string_list_sort(include);
 
-	for (i = 0; i < midx_pack_names_nr; i++) {
-		const char *pack_name = midx_pack_names[i];
+	for_each_string_list_item(item, midx_pack_names) {
+		const char *pack_name = item->string;
 
 		/*
 		 * Determine whether or not each MIDX'd pack from the existing
@@ -191,8 +190,7 @@ static int midx_has_unknown_packs(char **midx_pack_names,
 
 static void midx_included_packs(struct string_list *include,
 				struct existing_packs *existing,
-				char **midx_pack_names,
-				size_t midx_pack_names_nr,
+				struct string_list *midx_pack_names,
 				struct string_list *names,
 				struct pack_geometry *geometry)
 {
@@ -247,8 +245,8 @@ static void midx_included_packs(struct string_list *include,
 	}
 
 	if (midx_must_contain_cruft ||
-	    midx_has_unknown_packs(midx_pack_names, midx_pack_names_nr,
-				   include, geometry, existing)) {
+	    midx_has_unknown_packs(midx_pack_names, include, geometry,
+				   existing)) {
 		/*
 		 * If there are one or more unknown pack(s) present (see
 		 * midx_has_unknown_packs() for what makes a pack
@@ -606,13 +604,12 @@ int cmd_repack(int argc,
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
 	struct string_list names = STRING_LIST_INIT_DUP;
+	struct string_list midx_pack_names = STRING_LIST_INIT_DUP;
 	struct existing_packs existing = EXISTING_PACKS_INIT;
 	struct pack_geometry geometry = { 0 };
 	struct tempfile *refs_snapshot = NULL;
 	int i, ret;
 	int show_progress;
-	char **midx_pack_names = NULL;
-	size_t midx_pack_names_nr = 0;
 
 	/* variables to be filled by option parsing */
 	struct repack_config_ctx config_ctx;
@@ -985,13 +982,12 @@ int cmd_repack(int argc,
 		struct multi_pack_index *m =
 			get_multi_pack_index(repo->objects->sources);
 
-		ALLOC_ARRAY(midx_pack_names,
-			    m->num_packs + m->num_packs_in_base);
-
-		for (; m; m = m->base_midx)
-			for (uint32_t i = 0; i < m->num_packs; i++)
-				midx_pack_names[midx_pack_names_nr++] =
-					xstrdup(m->pack_names[i]);
+		for (; m; m = m->base_midx) {
+			for (uint32_t i = 0; i < m->num_packs; i++) {
+				string_list_append(&midx_pack_names,
+						   m->pack_names[i]);
+			}
+		}
 	}
 
 	close_object_store(repo->objects);
@@ -1019,8 +1015,8 @@ int cmd_repack(int argc,
 			.write_bitmaps = write_bitmaps > 0,
 			.midx_must_contain_cruft = midx_must_contain_cruft
 		};
-		midx_included_packs(&include, &existing, midx_pack_names,
-				    midx_pack_names_nr, &names, &geometry);
+		midx_included_packs(&include, &existing, &midx_pack_names,
+				    &names, &geometry);
 
 		ret = write_midx_included_packs(&opts);
 
@@ -1067,11 +1063,9 @@ int cmd_repack(int argc,
 cleanup:
 	string_list_clear(&keep_pack_list, 0);
 	string_list_clear(&names, 1);
+	string_list_clear(&midx_pack_names, 0);
 	existing_packs_release(&existing);
 	pack_geometry_release(&geometry);
-	for (size_t i = 0; i < midx_pack_names_nr; i++)
-		free(midx_pack_names[i]);
-	free(midx_pack_names);
 	pack_objects_args_release(&po_args);
 	pack_objects_args_release(&cruft_po_args);
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:28 UTC · re: Taylor Blau · lore

[PATCH v2 34/49] repack: keep track of MIDX pack names using existing_packs

Instead of storing the list of MIDX pack names separately, let's inline it into the existing_packs struct, further reducing the number of parameters we have to pass around.

This amounts to adding a new string_list to the existing_packs struct, and populating it via `existing_packs_collect()`. This is fairly straightforward to do, since we are already looping over all packs, all we need to do is:

    if (p->multi_pack_index)
        string_list_append(&existing->midx_packs, pack_basename(p));

Note, however, that this check *must* come before other conditions where we discard and do not keep track of a pack, including the condition "if (!p->pack_local)" immediately below. This is because the existing routine which collects MIDX pack names does so blindly, and does not discard, for example, non-local packs.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 26 ++++----------------------
 repack.c         |  5 +++++
 repack.h         |  1 +
 3 files changed, 10 insertions(+), 22 deletions(-)
Show changes to 3 files +10 −22

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index dda533f171..a57a14ef60 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -118,8 +118,7 @@ struct repack_write_midx_opts {
 	int midx_must_contain_cruft;
 };
 
-static int midx_has_unknown_packs(struct string_list *midx_pack_names,
-				  struct string_list *include,
+static int midx_has_unknown_packs(struct string_list *include,
 				  struct pack_geometry *geometry,
 				  struct existing_packs *existing)
 {
@@ -127,7 +126,7 @@ static int midx_has_unknown_packs(struct string_list *midx_pack_names,
 
 	string_list_sort(include);
 
-	for_each_string_list_item(item, midx_pack_names) {
+	for_each_string_list_item(item, &existing->midx_packs) {
 		const char *pack_name = item->string;
 
 		/*
@@ -190,7 +189,6 @@ static int midx_has_unknown_packs(struct string_list *midx_pack_names,
 
 static void midx_included_packs(struct string_list *include,
 				struct existing_packs *existing,
-				struct string_list *midx_pack_names,
 				struct string_list *names,
 				struct pack_geometry *geometry)
 {
@@ -245,8 +243,7 @@ static void midx_included_packs(struct string_list *include,
 	}
 
 	if (midx_must_contain_cruft ||
-	    midx_has_unknown_packs(midx_pack_names, include, geometry,
-				   existing)) {
+	    midx_has_unknown_packs(include, geometry, existing)) {
 		/*
 		 * If there are one or more unknown pack(s) present (see
 		 * midx_has_unknown_packs() for what makes a pack
@@ -604,7 +601,6 @@ int cmd_repack(int argc,
 	struct child_process cmd = CHILD_PROCESS_INIT;
 	struct string_list_item *item;
 	struct string_list names = STRING_LIST_INIT_DUP;
-	struct string_list midx_pack_names = STRING_LIST_INIT_DUP;
 	struct existing_packs existing = EXISTING_PACKS_INIT;
 	struct pack_geometry geometry = { 0 };
 	struct tempfile *refs_snapshot = NULL;
@@ -978,18 +974,6 @@ int cmd_repack(int argc,
 
 	string_list_sort(&names);
 
-	if (get_multi_pack_index(repo->objects->sources)) {
-		struct multi_pack_index *m =
-			get_multi_pack_index(repo->objects->sources);
-
-		for (; m; m = m->base_midx) {
-			for (uint32_t i = 0; i < m->num_packs; i++) {
-				string_list_append(&midx_pack_names,
-						   m->pack_names[i]);
-			}
-		}
-	}
-
 	close_object_store(repo->objects);
 
 	/*
@@ -1015,8 +999,7 @@ int cmd_repack(int argc,
 			.write_bitmaps = write_bitmaps > 0,
 			.midx_must_contain_cruft = midx_must_contain_cruft
 		};
-		midx_included_packs(&include, &existing, &midx_pack_names,
-				    &names, &geometry);
+		midx_included_packs(&include, &existing, &names, &geometry);
 
 		ret = write_midx_included_packs(&opts);
 
@@ -1063,7 +1046,6 @@ int cmd_repack(int argc,
 cleanup:
 	string_list_clear(&keep_pack_list, 0);
 	string_list_clear(&names, 1);
-	string_list_clear(&midx_pack_names, 0);
 	existing_packs_release(&existing);
 	pack_geometry_release(&geometry);
 	pack_objects_args_release(&po_args);
diff --git a/repack.c b/repack.c
index d8afdd352d..1d485e0112 100644
--- a/repack.c
+++ b/repack.c
@@ -80,6 +80,9 @@ void existing_packs_collect(struct existing_packs *existing,
 		size_t i;
 		const char *base;
 
+		if (p->multi_pack_index)
+			string_list_append(&existing->midx_packs,
+					    pack_basename(p));
 		if (!p->pack_local)
 			continue;
 
@@ -104,6 +107,7 @@ void existing_packs_collect(struct existing_packs *existing,
 	string_list_sort(&existing->kept_packs);
 	string_list_sort(&existing->non_kept_packs);
 	string_list_sort(&existing->cruft_packs);
+	string_list_sort(&existing->midx_packs);
 	strbuf_release(&buf);
 }
 
@@ -220,6 +224,7 @@ void existing_packs_release(struct existing_packs *existing)
 	string_list_clear(&existing->kept_packs, 0);
 	string_list_clear(&existing->non_kept_packs, 0);
 	string_list_clear(&existing->cruft_packs, 0);
+	string_list_clear(&existing->midx_packs, 0);
 }
 
 static struct {
diff --git a/repack.h b/repack.h
index 803e129224..6aa5b4e0f0 100644
--- a/repack.h
+++ b/repack.h
@@ -40,6 +40,7 @@ struct existing_packs {
 	struct string_list kept_packs;
 	struct string_list non_kept_packs;
 	struct string_list cruft_packs;
+	struct string_list midx_packs;
 };
 
 #define EXISTING_PACKS_INIT { \
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 35/49] builtin/repack.c: reorder `remove_redundant_bitmaps()`

The next commit will inline the call to `remove_redundant_bitmaps()` into `write_midx_included_packs()`. Reorder these two functions to avoid a forward declaration to `remove_redundant_bitmaps()`.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 58 ++++++++++++++++++++++++------------------------
 1 file changed, 29 insertions(+), 29 deletions(-)
Show changes to builtin/repack.c +29 −29
diff --git a/builtin/repack.c b/builtin/repack.c
index a57a14ef60..865e0af039 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -296,6 +296,35 @@ static void midx_included_packs(struct string_list *include,
 	strbuf_release(&buf);
 }
 
+static void remove_redundant_bitmaps(struct string_list *include,
+				     const char *packdir)
+{
+	struct strbuf path = STRBUF_INIT;
+	struct string_list_item *item;
+	size_t packdir_len;
+
+	strbuf_addstr(&path, packdir);
+	strbuf_addch(&path, '/');
+	packdir_len = path.len;
+
+	/*
+	 * Remove any pack bitmaps corresponding to packs which are now
+	 * included in the MIDX.
+	 */
+	for_each_string_list_item(item, include) {
+		strbuf_addstr(&path, item->string);
+		strbuf_strip_suffix(&path, ".idx");
+		strbuf_addstr(&path, ".bitmap");
+
+		if (unlink(path.buf) && errno != ENOENT)
+			warning_errno(_("could not remove stale bitmap: %s"),
+				      path.buf);
+
+		strbuf_setlen(&path, packdir_len);
+	}
+	strbuf_release(&path);
+}
+
 static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
@@ -373,35 +402,6 @@ static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 	return finish_command(&cmd);
 }
 
-static void remove_redundant_bitmaps(struct string_list *include,
-				     const char *packdir)
-{
-	struct strbuf path = STRBUF_INIT;
-	struct string_list_item *item;
-	size_t packdir_len;
-
-	strbuf_addstr(&path, packdir);
-	strbuf_addch(&path, '/');
-	packdir_len = path.len;
-
-	/*
-	 * Remove any pack bitmaps corresponding to packs which are now
-	 * included in the MIDX.
-	 */
-	for_each_string_list_item(item, include) {
-		strbuf_addstr(&path, item->string);
-		strbuf_strip_suffix(&path, ".idx");
-		strbuf_addstr(&path, ".bitmap");
-
-		if (unlink(path.buf) && errno != ENOENT)
-			warning_errno(_("could not remove stale bitmap: %s"),
-				      path.buf);
-
-		strbuf_setlen(&path, packdir_len);
-	}
-	strbuf_release(&path);
-}
-
 static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 				   struct child_process *cmd,
 				   struct string_list *names,
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 36/49] builtin/repack.c: inline `remove_redundant_bitmaps()`

After writing a new MIDX, the repack command removes any bitmaps belonging to packs which were written into the MIDX.

This is currently done in a separate function outside of `write_midx_included_packs()`, which forces the caller to keep track of the set of packs written into the MIDX.

Prepare to no longer require the caller to keep track of such information by inlining the clean-up into `write_midx_included_packs()`. Future commits will make the caller oblivious to the set of packs included in the MIDX altogether.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)
Show changes to builtin/repack.c +8 −7
diff --git a/builtin/repack.c b/builtin/repack.c
index 865e0af039..271c869268 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -331,10 +331,10 @@ static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 	struct string_list_item *item;
 	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
 	FILE *in;
-	int ret;
+	int ret = 0;
 
 	if (!opts->include->nr)
-		return 0;
+		goto done;
 
 	cmd.in = -1;
 	cmd.git_cmd = 1;
@@ -392,14 +392,18 @@ static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 
 	ret = start_command(&cmd);
 	if (ret)
-		return ret;
+		goto done;
 
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, opts->include)
 		fprintf(in, "%s\n", item->string);
 	fclose(in);
 
-	return finish_command(&cmd);
+	ret = finish_command(&cmd);
+done:
+	if (!ret && opts->write_bitmaps)
+		remove_redundant_bitmaps(opts->include, opts->packdir);
+	return ret;
 }
 
 static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
@@ -1003,9 +1007,6 @@ int cmd_repack(int argc,
 
 		ret = write_midx_included_packs(&opts);
 
-		if (!ret && write_bitmaps)
-			remove_redundant_bitmaps(&include, opts.packdir);
-
 		string_list_clear(&include, 0);
 
 		if (ret)
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 37/49] builtin/repack.c: pass `repack_write_midx_opts` to `midx_included_packs`

Instead of passing individual parameters (in this case, "existing", "names", and "geometry") to `midx_included_packs()`, pass a pointer to a `repack_write_midx_opts` structure instead.

Besides reducing the number of parameters necessary to call the `midx_included_packs` function, this refactoring sets us up nicely to inline the call to `midx_included_packs()` into `write_midx_included_packs()`, thus making the caller (in this case, `cmd_repack()`) oblivious to the set of packs being written into the MIDX.

In order to do this, `repack_write_midx_opts` has to keep track of the set of existing packs, so add an additional field to point to that set.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)
Show changes to builtin/repack.c +8 −5
diff --git a/builtin/repack.c b/builtin/repack.c
index 271c869268..a518a2b2f3 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -108,6 +108,7 @@ static int repack_config(const char *var, const char *value,
 }
 
 struct repack_write_midx_opts {
+	struct existing_packs *existing;
 	struct string_list *include;
 	struct pack_geometry *geometry;
 	struct string_list *names;
@@ -188,10 +189,11 @@ static int midx_has_unknown_packs(struct string_list *include,
 }
 
 static void midx_included_packs(struct string_list *include,
-				struct existing_packs *existing,
-				struct string_list *names,
-				struct pack_geometry *geometry)
+				struct repack_write_midx_opts *opts)
 {
+	struct existing_packs *existing = opts->existing;
+	struct pack_geometry *geometry = opts->geometry;
+	struct string_list *names = opts->names;
 	struct string_list_item *item;
 	struct strbuf buf = STRBUF_INIT;
 
@@ -242,7 +244,7 @@ static void midx_included_packs(struct string_list *include,
 		}
 	}
 
-	if (midx_must_contain_cruft ||
+	if (opts->midx_must_contain_cruft ||
 	    midx_has_unknown_packs(include, geometry, existing)) {
 		/*
 		 * If there are one or more unknown pack(s) present (see
@@ -994,6 +996,7 @@ int cmd_repack(int argc,
 	if (write_midx) {
 		struct string_list include = STRING_LIST_INIT_DUP;
 		struct repack_write_midx_opts opts = {
+			.existing = &existing,
 			.include = &include,
 			.geometry = &geometry,
 			.names = &names,
@@ -1003,7 +1006,7 @@ int cmd_repack(int argc,
 			.write_bitmaps = write_bitmaps > 0,
 			.midx_must_contain_cruft = midx_must_contain_cruft
 		};
-		midx_included_packs(&include, &existing, &names, &geometry);
+		midx_included_packs(&include, &opts);
 
 		ret = write_midx_included_packs(&opts);
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 38/49] builtin/repack.c: inline packs within `write_midx_included_packs()`

To write a MIDX at the end of a repack operation, 'git repack' presently computes the set of packs to write into the MIDX, before invoking `write_midx_included_packs()` with a `string_list` containing those packs.

The logic for computing which packs are supposed to appear in the resulting MIDX is within `midx_included_packs()`, where it is aware of details like which cruft pack(s) were written/combined, if/how we did a geometric repack, etc.

Computing this list ourselves before providing it to the sole function to make use of that list `write_midx_included_packs()` is somewhat awkward. In the future, repack will learn how to write incremental MIDXs, which will use a very different pack selection routine.

Instead of doing something like:
    struct string_list included_packs = STRING_LIST_INIT_DUP;
    if (incremental) {
        midx_incremental_included_packs(&included_packs, ...):
        write_midx_incremental_included_packs(&included_packs, ...);
    } else {
        midx_included_packs(&included_packs, ...):
        write_midx_included_packs(&included_packs, ...);
    }

in the future, let's have each function that writes a MIDX be responsible for itself computing the list of included packs. Inline the declaration and initialization of `included_packs` into the `write_midx_included_packs()` function itself, and repeat that pattern in the future when we introduce new ways to write MIDXs.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 17 ++++++++---------
 1 file changed, 8 insertions(+), 9 deletions(-)
Show changes to builtin/repack.c +8 −9
diff --git a/builtin/repack.c b/builtin/repack.c
index a518a2b2f3..fad10be42a 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -109,7 +109,6 @@ static int repack_config(const char *var, const char *value,
 
 struct repack_write_midx_opts {
 	struct existing_packs *existing;
-	struct string_list *include;
 	struct pack_geometry *geometry;
 	struct string_list *names;
 	const char *refs_snapshot;
@@ -330,12 +329,14 @@ static void remove_redundant_bitmaps(struct string_list *include,
 static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 {
 	struct child_process cmd = CHILD_PROCESS_INIT;
+	struct string_list include = STRING_LIST_INIT_DUP;
 	struct string_list_item *item;
 	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
 	FILE *in;
 	int ret = 0;
 
-	if (!opts->include->nr)
+	midx_included_packs(&include, opts);
+	if (!include.nr)
 		goto done;
 
 	cmd.in = -1;
@@ -397,14 +398,17 @@ static int write_midx_included_packs(struct repack_write_midx_opts *opts)
 		goto done;
 
 	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, opts->include)
+	for_each_string_list_item(item, &include)
 		fprintf(in, "%s\n", item->string);
 	fclose(in);
 
 	ret = finish_command(&cmd);
 done:
 	if (!ret && opts->write_bitmaps)
-		remove_redundant_bitmaps(opts->include, opts->packdir);
+		remove_redundant_bitmaps(&include, opts->packdir);
+
+	string_list_clear(&include, 0);
+
 	return ret;
 }
 
@@ -994,10 +998,8 @@ int cmd_repack(int argc,
 		existing_packs_mark_for_deletion(&existing, &names);
 
 	if (write_midx) {
-		struct string_list include = STRING_LIST_INIT_DUP;
 		struct repack_write_midx_opts opts = {
 			.existing = &existing,
-			.include = &include,
 			.geometry = &geometry,
 			.names = &names,
 			.refs_snapshot = refs_snapshot ? get_tempfile_path(refs_snapshot) : NULL,
@@ -1006,12 +1008,9 @@ int cmd_repack(int argc,
 			.write_bitmaps = write_bitmaps > 0,
 			.midx_must_contain_cruft = midx_must_contain_cruft
 		};
-		midx_included_packs(&include, &opts);
 
 		ret = write_midx_included_packs(&opts);
 
-		string_list_clear(&include, 0);
-
 		if (ret)
 			goto cleanup;
 	}
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 39/49] repack: 'write_midx_included_packs' API from the builtin

Now that we have sufficiently cleaned up the write_midx_included_packs() function, we can move it (along with the struct repack_write_midx_opts) out of the builtin, and into the repack.h header.

Since this function (and the static ones that it depends on) are MIDX-specific details of the repacking process, move them to the repack-midx.c compilation unit instead of the general repack.c one.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 305 -----------------------------------------------
 repack-midx.c    | 295 +++++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  12 ++
 3 files changed, 307 insertions(+), 305 deletions(-)
Show changes to 3 files +307 −305

builtin/repack.c, repack-midx.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index fad10be42a..2f49a18283 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,311 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-struct repack_write_midx_opts {
-	struct existing_packs *existing;
-	struct pack_geometry *geometry;
-	struct string_list *names;
-	const char *refs_snapshot;
-	const char *packdir;
-	int show_progress;
-	int write_bitmaps;
-	int midx_must_contain_cruft;
-};
-
-static int midx_has_unknown_packs(struct string_list *include,
-				  struct pack_geometry *geometry,
-				  struct existing_packs *existing)
-{
-	struct string_list_item *item;
-
-	string_list_sort(include);
-
-	for_each_string_list_item(item, &existing->midx_packs) {
-		const char *pack_name = item->string;
-
-		/*
-		 * Determine whether or not each MIDX'd pack from the existing
-		 * MIDX (if any) is represented in the new MIDX. For each pack
-		 * in the MIDX, it must either be:
-		 *
-		 *  - In the "include" list of packs to be included in the new
-		 *    MIDX. Note this function is called before the include
-		 *    list is populated with any cruft pack(s).
-		 *
-		 *  - Below the geometric split line (if using pack geometry),
-		 *    indicating that the pack won't be included in the new
-		 *    MIDX, but its contents were rolled up as part of the
-		 *    geometric repack.
-		 *
-		 *  - In the existing non-kept packs list (if not using pack
-		 *    geometry), and marked as non-deleted.
-		 */
-		if (string_list_has_string(include, pack_name)) {
-			continue;
-		} else if (geometry) {
-			struct strbuf buf = STRBUF_INIT;
-			uint32_t j;
-
-			for (j = 0; j < geometry->split; j++) {
-				strbuf_reset(&buf);
-				strbuf_addstr(&buf, pack_basename(geometry->pack[j]));
-				strbuf_strip_suffix(&buf, ".pack");
-				strbuf_addstr(&buf, ".idx");
-
-				if (!strcmp(pack_name, buf.buf)) {
-					strbuf_release(&buf);
-					break;
-				}
-			}
-
-			strbuf_release(&buf);
-
-			if (j < geometry->split)
-				continue;
-		} else {
-			struct string_list_item *item;
-
-			item = string_list_lookup(&existing->non_kept_packs,
-						  pack_name);
-			if (item && !existing_pack_is_marked_for_deletion(item))
-				continue;
-		}
-
-		/*
-		 * If we got to this point, the MIDX includes some pack that we
-		 * don't know about.
-		 */
-		return 1;
-	}
-
-	return 0;
-}
-
-static void midx_included_packs(struct string_list *include,
-				struct repack_write_midx_opts *opts)
-{
-	struct existing_packs *existing = opts->existing;
-	struct pack_geometry *geometry = opts->geometry;
-	struct string_list *names = opts->names;
-	struct string_list_item *item;
-	struct strbuf buf = STRBUF_INIT;
-
-	for_each_string_list_item(item, &existing->kept_packs) {
-		strbuf_reset(&buf);
-		strbuf_addf(&buf, "%s.idx", item->string);
-		string_list_insert(include, buf.buf);
-	}
-
-	for_each_string_list_item(item, names) {
-		strbuf_reset(&buf);
-		strbuf_addf(&buf, "pack-%s.idx", item->string);
-		string_list_insert(include, buf.buf);
-	}
-
-	if (geometry->split_factor) {
-		uint32_t i;
-
-		for (i = geometry->split; i < geometry->pack_nr; i++) {
-			struct packed_git *p = geometry->pack[i];
-
-			/*
-			 * The multi-pack index never refers to packfiles part
-			 * of an alternate object database, so we skip these.
-			 * While git-multi-pack-index(1) would silently ignore
-			 * them anyway, this allows us to skip executing the
-			 * command completely when we have only non-local
-			 * packfiles.
-			 */
-			if (!p->pack_local)
-				continue;
-
-			strbuf_reset(&buf);
-			strbuf_addstr(&buf, pack_basename(p));
-			strbuf_strip_suffix(&buf, ".pack");
-			strbuf_addstr(&buf, ".idx");
-
-			string_list_insert(include, buf.buf);
-		}
-	} else {
-		for_each_string_list_item(item, &existing->non_kept_packs) {
-			if (existing_pack_is_marked_for_deletion(item))
-				continue;
-
-			strbuf_reset(&buf);
-			strbuf_addf(&buf, "%s.idx", item->string);
-			string_list_insert(include, buf.buf);
-		}
-	}
-
-	if (opts->midx_must_contain_cruft ||
-	    midx_has_unknown_packs(include, geometry, existing)) {
-		/*
-		 * If there are one or more unknown pack(s) present (see
-		 * midx_has_unknown_packs() for what makes a pack
-		 * "unknown") in the MIDX before the repack, keep them
-		 * as they may be required to form a reachability
-		 * closure if the MIDX is bitmapped.
-		 *
-		 * For example, a cruft pack can be required to form a
-		 * reachability closure if the MIDX is bitmapped and one
-		 * or more of the bitmap's selected commits reaches a
-		 * once-cruft object that was later made reachable.
-		 */
-		for_each_string_list_item(item, &existing->cruft_packs) {
-			/*
-			 * When doing a --geometric repack, there is no
-			 * need to check for deleted packs, since we're
-			 * by definition not doing an ALL_INTO_ONE
-			 * repack (hence no packs will be deleted).
-			 * Otherwise we must check for and exclude any
-			 * packs which are enqueued for deletion.
-			 *
-			 * So we could omit the conditional below in the
-			 * --geometric case, but doing so is unnecessary
-			 *  since no packs are marked as pending
-			 *  deletion (since we only call
-			 *  `existing_packs_mark_for_deletion()` when
-			 *  doing an all-into-one repack).
-			 */
-			if (existing_pack_is_marked_for_deletion(item))
-				continue;
-
-			strbuf_reset(&buf);
-			strbuf_addf(&buf, "%s.idx", item->string);
-			string_list_insert(include, buf.buf);
-		}
-	} else {
-		/*
-		 * Modern versions of Git (with the appropriate
-		 * configuration setting) will write new copies of
-		 * once-cruft objects when doing a --geometric repack.
-		 *
-		 * If the MIDX has no cruft pack, new packs written
-		 * during a --geometric repack will not rely on the
-		 * cruft pack to form a reachability closure, so we can
-		 * avoid including them in the MIDX in that case.
-		 */
-		;
-	}
-
-	strbuf_release(&buf);
-}
-
-static void remove_redundant_bitmaps(struct string_list *include,
-				     const char *packdir)
-{
-	struct strbuf path = STRBUF_INIT;
-	struct string_list_item *item;
-	size_t packdir_len;
-
-	strbuf_addstr(&path, packdir);
-	strbuf_addch(&path, '/');
-	packdir_len = path.len;
-
-	/*
-	 * Remove any pack bitmaps corresponding to packs which are now
-	 * included in the MIDX.
-	 */
-	for_each_string_list_item(item, include) {
-		strbuf_addstr(&path, item->string);
-		strbuf_strip_suffix(&path, ".idx");
-		strbuf_addstr(&path, ".bitmap");
-
-		if (unlink(path.buf) && errno != ENOENT)
-			warning_errno(_("could not remove stale bitmap: %s"),
-				      path.buf);
-
-		strbuf_setlen(&path, packdir_len);
-	}
-	strbuf_release(&path);
-}
-
-static int write_midx_included_packs(struct repack_write_midx_opts *opts)
-{
-	struct child_process cmd = CHILD_PROCESS_INIT;
-	struct string_list include = STRING_LIST_INIT_DUP;
-	struct string_list_item *item;
-	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
-	FILE *in;
-	int ret = 0;
-
-	midx_included_packs(&include, opts);
-	if (!include.nr)
-		goto done;
-
-	cmd.in = -1;
-	cmd.git_cmd = 1;
-
-	strvec_push(&cmd.args, "multi-pack-index");
-	strvec_pushl(&cmd.args, "write", "--stdin-packs", NULL);
-
-	if (opts->show_progress)
-		strvec_push(&cmd.args, "--progress");
-	else
-		strvec_push(&cmd.args, "--no-progress");
-
-	if (opts->write_bitmaps)
-		strvec_push(&cmd.args, "--bitmap");
-
-	if (preferred)
-		strvec_pushf(&cmd.args, "--preferred-pack=%s",
-			     pack_basename(preferred));
-	else if (opts->names->nr) {
-		/* The largest pack was repacked, meaning that either
-		 * one or two packs exist depending on whether the
-		 * repository has a cruft pack or not.
-		 *
-		 * Select the non-cruft one as preferred to encourage
-		 * pack-reuse among packs containing reachable objects
-		 * over unreachable ones.
-		 *
-		 * (Note we could write multiple packs here if
-		 * `--max-pack-size` was given, but any one of them
-		 * will suffice, so pick the first one.)
-		 */
-		for_each_string_list_item(item, opts->names) {
-			struct generated_pack *pack = item->util;
-			if (generated_pack_has_ext(pack, ".mtimes"))
-				continue;
-
-			strvec_pushf(&cmd.args, "--preferred-pack=pack-%s.pack",
-				     item->string);
-			break;
-		}
-	} else {
-		/*
-		 * No packs were kept, and no packs were written. The
-		 * only thing remaining are .keep packs (unless
-		 * --pack-kept-objects was given).
-		 *
-		 * Set the `--preferred-pack` arbitrarily here.
-		 */
-		;
-	}
-
-	if (opts->refs_snapshot)
-		strvec_pushf(&cmd.args, "--refs-snapshot=%s",
-			     opts->refs_snapshot);
-
-	ret = start_command(&cmd);
-	if (ret)
-		goto done;
-
-	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, &include)
-		fprintf(in, "%s\n", item->string);
-	fclose(in);
-
-	ret = finish_command(&cmd);
-done:
-	if (!ret && opts->write_bitmaps)
-		remove_redundant_bitmaps(&include, opts->packdir);
-
-	string_list_clear(&include, 0);
-
-	return ret;
-}
-
 static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 				   struct child_process *cmd,
 				   struct string_list *names,
diff --git a/repack-midx.c b/repack-midx.c
index 354df729a5..6f6202c5bc 100644
--- a/repack-midx.c
+++ b/repack-midx.c
@@ -6,6 +6,7 @@
 #include "oidset.h"
 #include "pack-bitmap.h"
 #include "refs.h"
+#include "run-command.h"
 #include "tempfile.h"
 
 struct midx_snapshot_ref_data {
@@ -75,3 +76,297 @@ void midx_snapshot_refs(struct repository *repo, struct tempfile *f)
 
 	oidset_clear(&data.seen);
 }
+
+static int midx_has_unknown_packs(struct string_list *include,
+				  struct pack_geometry *geometry,
+				  struct existing_packs *existing)
+{
+	struct string_list_item *item;
+
+	string_list_sort(include);
+
+	for_each_string_list_item(item, &existing->midx_packs) {
+		const char *pack_name = item->string;
+
+		/*
+		 * Determine whether or not each MIDX'd pack from the existing
+		 * MIDX (if any) is represented in the new MIDX. For each pack
+		 * in the MIDX, it must either be:
+		 *
+		 *  - In the "include" list of packs to be included in the new
+		 *    MIDX. Note this function is called before the include
+		 *    list is populated with any cruft pack(s).
+		 *
+		 *  - Below the geometric split line (if using pack geometry),
+		 *    indicating that the pack won't be included in the new
+		 *    MIDX, but its contents were rolled up as part of the
+		 *    geometric repack.
+		 *
+		 *  - In the existing non-kept packs list (if not using pack
+		 *    geometry), and marked as non-deleted.
+		 */
+		if (string_list_has_string(include, pack_name)) {
+			continue;
+		} else if (geometry) {
+			struct strbuf buf = STRBUF_INIT;
+			uint32_t j;
+
+			for (j = 0; j < geometry->split; j++) {
+				strbuf_reset(&buf);
+				strbuf_addstr(&buf, pack_basename(geometry->pack[j]));
+				strbuf_strip_suffix(&buf, ".pack");
+				strbuf_addstr(&buf, ".idx");
+
+				if (!strcmp(pack_name, buf.buf)) {
+					strbuf_release(&buf);
+					break;
+				}
+			}
+
+			strbuf_release(&buf);
+
+			if (j < geometry->split)
+				continue;
+		} else {
+			struct string_list_item *item;
+
+			item = string_list_lookup(&existing->non_kept_packs,
+						  pack_name);
+			if (item && !existing_pack_is_marked_for_deletion(item))
+				continue;
+		}
+
+		/*
+		 * If we got to this point, the MIDX includes some pack that we
+		 * don't know about.
+		 */
+		return 1;
+	}
+
+	return 0;
+}
+
+static void midx_included_packs(struct string_list *include,
+				struct repack_write_midx_opts *opts)
+{
+	struct existing_packs *existing = opts->existing;
+	struct pack_geometry *geometry = opts->geometry;
+	struct string_list *names = opts->names;
+	struct string_list_item *item;
+	struct strbuf buf = STRBUF_INIT;
+
+	for_each_string_list_item(item, &existing->kept_packs) {
+		strbuf_reset(&buf);
+		strbuf_addf(&buf, "%s.idx", item->string);
+		string_list_insert(include, buf.buf);
+	}
+
+	for_each_string_list_item(item, names) {
+		strbuf_reset(&buf);
+		strbuf_addf(&buf, "pack-%s.idx", item->string);
+		string_list_insert(include, buf.buf);
+	}
+
+	if (geometry->split_factor) {
+		uint32_t i;
+
+		for (i = geometry->split; i < geometry->pack_nr; i++) {
+			struct packed_git *p = geometry->pack[i];
+
+			/*
+			 * The multi-pack index never refers to packfiles part
+			 * of an alternate object database, so we skip these.
+			 * While git-multi-pack-index(1) would silently ignore
+			 * them anyway, this allows us to skip executing the
+			 * command completely when we have only non-local
+			 * packfiles.
+			 */
+			if (!p->pack_local)
+				continue;
+
+			strbuf_reset(&buf);
+			strbuf_addstr(&buf, pack_basename(p));
+			strbuf_strip_suffix(&buf, ".pack");
+			strbuf_addstr(&buf, ".idx");
+
+			string_list_insert(include, buf.buf);
+		}
+	} else {
+		for_each_string_list_item(item, &existing->non_kept_packs) {
+			if (existing_pack_is_marked_for_deletion(item))
+				continue;
+
+			strbuf_reset(&buf);
+			strbuf_addf(&buf, "%s.idx", item->string);
+			string_list_insert(include, buf.buf);
+		}
+	}
+
+	if (opts->midx_must_contain_cruft ||
+	    midx_has_unknown_packs(include, geometry, existing)) {
+		/*
+		 * If there are one or more unknown pack(s) present (see
+		 * midx_has_unknown_packs() for what makes a pack
+		 * "unknown") in the MIDX before the repack, keep them
+		 * as they may be required to form a reachability
+		 * closure if the MIDX is bitmapped.
+		 *
+		 * For example, a cruft pack can be required to form a
+		 * reachability closure if the MIDX is bitmapped and one
+		 * or more of the bitmap's selected commits reaches a
+		 * once-cruft object that was later made reachable.
+		 */
+		for_each_string_list_item(item, &existing->cruft_packs) {
+			/*
+			 * When doing a --geometric repack, there is no
+			 * need to check for deleted packs, since we're
+			 * by definition not doing an ALL_INTO_ONE
+			 * repack (hence no packs will be deleted).
+			 * Otherwise we must check for and exclude any
+			 * packs which are enqueued for deletion.
+			 *
+			 * So we could omit the conditional below in the
+			 * --geometric case, but doing so is unnecessary
+			 *  since no packs are marked as pending
+			 *  deletion (since we only call
+			 *  `existing_packs_mark_for_deletion()` when
+			 *  doing an all-into-one repack).
+			 */
+			if (existing_pack_is_marked_for_deletion(item))
+				continue;
+
+			strbuf_reset(&buf);
+			strbuf_addf(&buf, "%s.idx", item->string);
+			string_list_insert(include, buf.buf);
+		}
+	} else {
+		/*
+		 * Modern versions of Git (with the appropriate
+		 * configuration setting) will write new copies of
+		 * once-cruft objects when doing a --geometric repack.
+		 *
+		 * If the MIDX has no cruft pack, new packs written
+		 * during a --geometric repack will not rely on the
+		 * cruft pack to form a reachability closure, so we can
+		 * avoid including them in the MIDX in that case.
+		 */
+		;
+	}
+
+	strbuf_release(&buf);
+}
+
+static void remove_redundant_bitmaps(struct string_list *include,
+				     const char *packdir)
+{
+	struct strbuf path = STRBUF_INIT;
+	struct string_list_item *item;
+	size_t packdir_len;
+
+	strbuf_addstr(&path, packdir);
+	strbuf_addch(&path, '/');
+	packdir_len = path.len;
+
+	/*
+	 * Remove any pack bitmaps corresponding to packs which are now
+	 * included in the MIDX.
+	 */
+	for_each_string_list_item(item, include) {
+		strbuf_addstr(&path, item->string);
+		strbuf_strip_suffix(&path, ".idx");
+		strbuf_addstr(&path, ".bitmap");
+
+		if (unlink(path.buf) && errno != ENOENT)
+			warning_errno(_("could not remove stale bitmap: %s"),
+				      path.buf);
+
+		strbuf_setlen(&path, packdir_len);
+	}
+	strbuf_release(&path);
+}
+
+int write_midx_included_packs(struct repack_write_midx_opts *opts)
+{
+	struct child_process cmd = CHILD_PROCESS_INIT;
+	struct string_list include = STRING_LIST_INIT_DUP;
+	struct string_list_item *item;
+	struct packed_git *preferred = pack_geometry_preferred_pack(opts->geometry);
+	FILE *in;
+	int ret = 0;
+
+	midx_included_packs(&include, opts);
+	if (!include.nr)
+		goto done;
+
+	cmd.in = -1;
+	cmd.git_cmd = 1;
+
+	strvec_push(&cmd.args, "multi-pack-index");
+	strvec_pushl(&cmd.args, "write", "--stdin-packs", NULL);
+
+	if (opts->show_progress)
+		strvec_push(&cmd.args, "--progress");
+	else
+		strvec_push(&cmd.args, "--no-progress");
+
+	if (opts->write_bitmaps)
+		strvec_push(&cmd.args, "--bitmap");
+
+	if (preferred)
+		strvec_pushf(&cmd.args, "--preferred-pack=%s",
+			     pack_basename(preferred));
+	else if (opts->names->nr) {
+		/* The largest pack was repacked, meaning that either
+		 * one or two packs exist depending on whether the
+		 * repository has a cruft pack or not.
+		 *
+		 * Select the non-cruft one as preferred to encourage
+		 * pack-reuse among packs containing reachable objects
+		 * over unreachable ones.
+		 *
+		 * (Note we could write multiple packs here if
+		 * `--max-pack-size` was given, but any one of them
+		 * will suffice, so pick the first one.)
+		 */
+		for_each_string_list_item(item, opts->names) {
+			struct generated_pack *pack = item->util;
+			if (generated_pack_has_ext(pack, ".mtimes"))
+				continue;
+
+			strvec_pushf(&cmd.args, "--preferred-pack=pack-%s.pack",
+				     item->string);
+			break;
+		}
+	} else {
+		/*
+		 * No packs were kept, and no packs were written. The
+		 * only thing remaining are .keep packs (unless
+		 * --pack-kept-objects was given).
+		 *
+		 * Set the `--preferred-pack` arbitrarily here.
+		 */
+		;
+	}
+
+	if (opts->refs_snapshot)
+		strvec_pushf(&cmd.args, "--refs-snapshot=%s",
+			     opts->refs_snapshot);
+
+	ret = start_command(&cmd);
+	if (ret)
+		goto done;
+
+	in = xfdopen(cmd.in, "w");
+	for_each_string_list_item(item, &include)
+		fprintf(in, "%s\n", item->string);
+	fclose(in);
+
+	ret = finish_command(&cmd);
+done:
+	if (!ret && opts->write_bitmaps)
+		remove_redundant_bitmaps(&include, opts->packdir);
+
+	string_list_clear(&include, 0);
+
+	return ret;
+}
diff --git a/repack.h b/repack.h
index 6aa5b4e0f0..25a31ac0a0 100644
--- a/repack.h
+++ b/repack.h
@@ -101,6 +101,18 @@ void pack_geometry_release(struct pack_geometry *geometry);
 
 struct tempfile;
 
+struct repack_write_midx_opts {
+	struct existing_packs *existing;
+	struct pack_geometry *geometry;
+	struct string_list *names;
+	const char *refs_snapshot;
+	const char *packdir;
+	int show_progress;
+	int write_bitmaps;
+	int midx_must_contain_cruft;
+};
+
 void midx_snapshot_refs(struct repository *repo, struct tempfile *f);
+int write_midx_included_packs(struct repack_write_midx_opts *opts);
 
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 40/49] builtin/repack.c: introduce `struct write_pack_opts`

There are various functions within the 'repack' builtin which are responsible for writing different kinds of packs. They include:

 - `static int write_filtered_pack(...)`
 - `static int write_cruft_pack(...)`

as well as the function `finish_pack_objects_cmd()`, which is responsible for finalizing a new pack write, and recording the checksum of its contents in the 'names' list.

Both of these `write_` functions have a few things in common. They both take a pointer to the 'pack_objects_args' struct, as well as a pair of character pointers for `destination` and `pack_prefix`.

Instead of repeating those arguments for each function, let's extract an options struct called "write_pack_opts" which has these three parameters as member fields. While we're at it, add fields for "packdir," and "packtmp", both of which are static variables within the builtin, and need to be read from within these two functions.

This will shorten the list of parameters that callers have to provide to `write_filtered_pack()`, avoid ambiguity when passing multiple variables of the same type, and provide a unified interface for the two functions mentioned earlier.

(Note that "pack_prefix" can be derived on the fly as a function of "packdir" and "packtmp", making it unnecessary to store "pack_prefix" explicitly. This commit ignores that potential cleanup in the name of doing as few things as possible, but a later commit will make that change.)

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 28 +++++++++++++++-------------
 repack.h         |  8 ++++++++
 2 files changed, 23 insertions(+), 13 deletions(-)
Show changes to 2 files +23 −13

builtin/repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 2f49a18283..45ce469898 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -138,9 +138,7 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 	return finish_command(cmd);
 }
 
-static int write_filtered_pack(const struct pack_objects_args *args,
-			       const char *destination,
-			       const char *pack_prefix,
+static int write_filtered_pack(const struct write_pack_opts *opts,
 			       struct existing_packs *existing,
 			       struct string_list *names)
 {
@@ -150,9 +148,9 @@ static int write_filtered_pack(const struct pack_objects_args *args,
 	int ret;
 	const char *caret;
 	const char *scratch;
-	int local = skip_prefix(destination, packdir, &scratch);
+	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
 
-	prepare_pack_objects(&cmd, args, destination);
+	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
 
 	strvec_push(&cmd.args, "--stdin-packs");
 
@@ -175,7 +173,7 @@ static int write_filtered_pack(const struct pack_objects_args *args,
 	 */
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, names)
-		fprintf(in, "^%s-%s.pack\n", pack_prefix, item->string);
+		fprintf(in, "^%s-%s.pack\n", opts->pack_prefix, item->string);
 	for_each_string_list_item(item, &existing->non_kept_packs)
 		fprintf(in, "%s.pack\n", item->string);
 	for_each_string_list_item(item, &existing->cruft_packs)
@@ -665,14 +663,18 @@ int cmd_repack(int argc,
 	}
 
 	if (po_args.filter_options.choice) {
-		if (!filter_to)
-			filter_to = packtmp;
+		struct write_pack_opts opts = {
+			.po_args = &po_args,
+			.destination = filter_to,
+			.pack_prefix = find_pack_prefix(packdir, packtmp),
+			.packdir = packdir,
+			.packtmp = packtmp,
+		};
 
-		ret = write_filtered_pack(&po_args,
-					  filter_to,
-					  find_pack_prefix(packdir, packtmp),
-					  &existing,
-					  &names);
+		if (!opts.destination)
+			opts.destination = packtmp;
+
+		ret = write_filtered_pack(&opts, &existing, &names);
 		if (ret)
 			goto cleanup;
 	}
diff --git a/repack.h b/repack.h
index 25a31ac0a0..6ef503f623 100644
--- a/repack.h
+++ b/repack.h
@@ -32,6 +32,14 @@ void pack_objects_args_release(struct pack_objects_args *args);
 void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 				  const char *base_name);
 
+struct write_pack_opts {
+	struct pack_objects_args *po_args;
+	const char *destination;
+	const char *pack_prefix;
+	const char *packdir;
+	const char *packtmp;
+};
+
 struct repository;
 struct packed_git;
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 41/49] builtin/repack.c: use `write_pack_opts` within `write_cruft_pack()`

Similar to the changes made in the previous commit to `write_filtered_pack()`, teach `write_cruft_pack()` to take a `write_pack_opts` struct and use that where possible.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)
Show changes to builtin/repack.c +14 −13
diff --git a/builtin/repack.c b/builtin/repack.c
index 45ce469898..7295135ec2 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -221,9 +221,7 @@ static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
 	strbuf_release(&buf);
 }
 
-static int write_cruft_pack(const struct pack_objects_args *args,
-			    const char *destination,
-			    const char *pack_prefix,
+static int write_cruft_pack(const struct write_pack_opts *opts,
 			    const char *cruft_expiration,
 			    unsigned long combine_cruft_below_size,
 			    struct string_list *names,
@@ -234,9 +232,9 @@ static int write_cruft_pack(const struct pack_objects_args *args,
 	FILE *in;
 	int ret;
 	const char *scratch;
-	int local = skip_prefix(destination, packdir, &scratch);
+	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
 
-	prepare_pack_objects(&cmd, args, destination);
+	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
 
 	strvec_push(&cmd.args, "--cruft");
 	if (cruft_expiration)
@@ -267,7 +265,7 @@ static int write_cruft_pack(const struct pack_objects_args *args,
 	 */
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, names)
-		fprintf(in, "%s-%s.pack\n", pack_prefix, item->string);
+		fprintf(in, "%s-%s.pack\n", opts->pack_prefix, item->string);
 	if (combine_cruft_below_size && !cruft_expiration) {
 		combine_small_cruft_packs(in, combine_cruft_below_size,
 					  existing);
@@ -599,6 +597,13 @@ int cmd_repack(int argc,
 
 	if (pack_everything & PACK_CRUFT) {
 		const char *pack_prefix = find_pack_prefix(packdir, packtmp);
+		struct write_pack_opts opts = {
+			.po_args = &cruft_po_args,
+			.destination = packtmp,
+			.pack_prefix = pack_prefix,
+			.packtmp = packtmp,
+			.packdir = packdir,
+		};
 
 		if (!cruft_po_args.window)
 			cruft_po_args.window = xstrdup_or_null(po_args.window);
@@ -615,8 +620,7 @@ int cmd_repack(int argc,
 		cruft_po_args.quiet = po_args.quiet;
 		cruft_po_args.delta_base_offset = po_args.delta_base_offset;
 
-		ret = write_cruft_pack(&cruft_po_args, packtmp, pack_prefix,
-				       cruft_expiration,
+		ret = write_cruft_pack(&opts, cruft_expiration,
 				       combine_cruft_below_size, &names,
 				       &existing);
 		if (ret)
@@ -651,11 +655,8 @@ int cmd_repack(int argc,
 			 * pack, but rather removing all cruft packs from the
 			 * main repository regardless of size.
 			 */
-			ret = write_cruft_pack(&cruft_po_args, expire_to,
-					       pack_prefix,
-					       NULL,
-					       0ul,
-					       &names,
+			opts.destination = expire_to;
+			ret = write_cruft_pack(&opts, NULL, 0ul, &names,
 					       &existing);
 			if (ret)
 				goto cleanup;
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 42/49] repack: move `find_pack_prefix()` out of the builtin

Both callers within the repack builtin which call functions that take a 'write_pack_opts' structure have the following pattern:

    struct write_pack_opts opts = {
        .packdir = packdir,
        .packtmp = packtmp,
        .pack_prefix = find_pack_prefix(packdir, packtmp),
        /* ... */
    };
    int ret = write_some_kind_of_pack(&opts, /* ... */);

, but both "packdir" and "packtmp" are fields within the write_pack_opts struct itself!

Instead of also computing the pack_prefix ahead of time, let's have the callees compute it themselves by moving `find_pack_prefix()` out of the repack builtin, and have it take a write_pack_opts pointer instead of the "packdir" and "packtmp" fields directly.

This avoids the callers having to do some prep work that is common between the two of them, but also avoids the potential pitfall of accidentally writing:

    .pack_prefix = find_pack_prefix(packtmp, packdir),
(which is well-typed) when the caller meant to instead write:
    .pack_prefix = find_pack_prefix(packdir, packtmp),
Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 20 ++++----------------
 repack.c         | 11 +++++++++++
 repack.h         |  3 ++-
 3 files changed, 17 insertions(+), 17 deletions(-)
Show changes to 3 files +17 −17

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 7295135ec2..b21799c650 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -149,6 +149,7 @@ static int write_filtered_pack(const struct write_pack_opts *opts,
 	const char *caret;
 	const char *scratch;
 	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
+	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
 
@@ -173,7 +174,7 @@ static int write_filtered_pack(const struct write_pack_opts *opts,
 	 */
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, names)
-		fprintf(in, "^%s-%s.pack\n", opts->pack_prefix, item->string);
+		fprintf(in, "^%s-%s.pack\n", pack_prefix, item->string);
 	for_each_string_list_item(item, &existing->non_kept_packs)
 		fprintf(in, "%s.pack\n", item->string);
 	for_each_string_list_item(item, &existing->cruft_packs)
@@ -233,6 +234,7 @@ static int write_cruft_pack(const struct write_pack_opts *opts,
 	int ret;
 	const char *scratch;
 	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
+	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
 
@@ -265,7 +267,7 @@ static int write_cruft_pack(const struct write_pack_opts *opts,
 	 */
 	in = xfdopen(cmd.in, "w");
 	for_each_string_list_item(item, names)
-		fprintf(in, "%s-%s.pack\n", opts->pack_prefix, item->string);
+		fprintf(in, "%s-%s.pack\n", pack_prefix, item->string);
 	if (combine_cruft_below_size && !cruft_expiration) {
 		combine_small_cruft_packs(in, combine_cruft_below_size,
 					  existing);
@@ -283,17 +285,6 @@ static int write_cruft_pack(const struct write_pack_opts *opts,
 				       local);
 }
 
-static const char *find_pack_prefix(const char *packdir, const char *packtmp)
-{
-	const char *pack_prefix;
-	if (!skip_prefix(packtmp, packdir, &pack_prefix))
-		die(_("pack prefix %s does not begin with objdir %s"),
-		    packtmp, packdir);
-	if (*pack_prefix == '/')
-		pack_prefix++;
-	return pack_prefix;
-}
-
 int cmd_repack(int argc,
 	       const char **argv,
 	       const char *prefix,
@@ -596,11 +587,9 @@ int cmd_repack(int argc,
 	}
 
 	if (pack_everything & PACK_CRUFT) {
-		const char *pack_prefix = find_pack_prefix(packdir, packtmp);
 		struct write_pack_opts opts = {
 			.po_args = &cruft_po_args,
 			.destination = packtmp,
-			.pack_prefix = pack_prefix,
 			.packtmp = packtmp,
 			.packdir = packdir,
 		};
@@ -667,7 +656,6 @@ int cmd_repack(int argc,
 		struct write_pack_opts opts = {
 			.po_args = &po_args,
 			.destination = filter_to,
-			.pack_prefix = find_pack_prefix(packdir, packtmp),
 			.packdir = packdir,
 			.packtmp = packtmp,
 		};
diff --git a/repack.c b/repack.c
index 1d485e0112..19fd1d6d5b 100644
--- a/repack.c
+++ b/repack.c
@@ -66,6 +66,17 @@ void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 	strbuf_release(&buf);
 }
 
+const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts)
+{
+	const char *pack_prefix;
+	if (!skip_prefix(opts->packtmp, opts->packdir, &pack_prefix))
+		die(_("pack prefix %s does not begin with objdir %s"),
+		    opts->packtmp, opts->packdir);
+	if (*pack_prefix == '/')
+		pack_prefix++;
+	return pack_prefix;
+}
+
 #define DELETE_PACK 1
 #define RETAIN_PACK 2
 
diff --git a/repack.h b/repack.h
index 6ef503f623..5852e2407f 100644
--- a/repack.h
+++ b/repack.h
@@ -35,11 +35,12 @@ void repack_remove_redundant_pack(struct repository *repo, const char *dir_name,
 struct write_pack_opts {
 	struct pack_objects_args *po_args;
 	const char *destination;
-	const char *pack_prefix;
 	const char *packdir;
 	const char *packtmp;
 };
 
+const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts);
+
 struct repository;
 struct packed_git;
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 43/49] repack: extract `write_pack_opts_is_local()`

Similar to the previous commit, the functions `write_cruft_pack()` and `write_filtered_pack()` both compute a "local" variable via the exact same mechanism:

    const char *scratch;
    int local = skip_prefix(opts->destination, opts->packdir, &scratch);

Not only does this cause us to repeat the same pair of lines, it also introduces an unnecessary "scratch" variable that is common between both functions.

Instead of repeating ourselves, let's extract that functionality into a new function in the repack.h API called "write_pack_opts_is_local()". That function takes a pointer to a "struct write_pack_opts" (which has as fields both "destination" and "packdir"), and can encapsulate the dangling "scratch" field.

Extract that function and make it visible within the repack.h API, and use it within both `write_cruft_pack()` and `write_filtered_pack()`. While we're at it, match our modern conventions by returning a "bool" instead of "int", and use `starts_with()` instead of `skip_prefix()` to avoid storing the dummy "scratch" variable.

The remaining duplication (that is, that both `write_cruft_pack()` and `write_filtered_pack()` still both call `write_pack_opts_is_local()`) will be addressed in the following commit.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 6 ++----
 repack.c         | 5 +++++
 repack.h         | 1 +
 3 files changed, 8 insertions(+), 4 deletions(-)
Show changes to 3 files +8 −4

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index b21799c650..d1449cfe13 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -147,8 +147,7 @@ static int write_filtered_pack(const struct write_pack_opts *opts,
 	FILE *in;
 	int ret;
 	const char *caret;
-	const char *scratch;
-	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
+	bool local = write_pack_opts_is_local(opts);
 	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
@@ -232,8 +231,7 @@ static int write_cruft_pack(const struct write_pack_opts *opts,
 	struct string_list_item *item;
 	FILE *in;
 	int ret;
-	const char *scratch;
-	int local = skip_prefix(opts->destination, opts->packdir, &scratch);
+	bool local = write_pack_opts_is_local(opts);
 	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
diff --git a/repack.c b/repack.c
index 19fd1d6d5b..d2ee9f2460 100644
--- a/repack.c
+++ b/repack.c
@@ -77,6 +77,11 @@ const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts)
 	return pack_prefix;
 }
 
+bool write_pack_opts_is_local(const struct write_pack_opts *opts)
+{
+	return starts_with(opts->destination, opts->packdir);
+}
+
 #define DELETE_PACK 1
 #define RETAIN_PACK 2
 
diff --git a/repack.h b/repack.h
index 5852e2407f..26d1954ae2 100644
--- a/repack.h
+++ b/repack.h
@@ -40,6 +40,7 @@ struct write_pack_opts {
 };
 
 const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts);
+bool write_pack_opts_is_local(const struct write_pack_opts *opts);
 
 struct repository;
 struct packed_git;
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 44/49] builtin/repack.c: pass `write_pack_opts` to `finish_pack_objects_cmd()`

To prepare to move the `finish_pack_objects_cmd()` function out of the builtin and into the repack.h API, there are a couple of things we need to do first:

 - First, let's take advantage of `write_pack_opts_is_local()` function
   introduced in the previous commit instead of passing "local"
   explicitly.
 - Let's also avoid referring to the static 'packtmp' field within
   builtin/repack.c by instead accessing it through the write_pack_opts
   argument.

There are three callers which need to adjust themselves in order to account for this change. The callers which reside in write_cruft_pack() and write_filtered_pack() both already have an "opts" in scope, so they can pass it through transparently.

The other call (at the bottom of `cmd_repack()`) needs to initialize its own write_pack_opts to pass the necessary fields over to the direct call to `finish_pack_objects_cmd()`.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 32 ++++++++++++++++++++------------
 1 file changed, 20 insertions(+), 12 deletions(-)
Show changes to builtin/repack.c +20 −12
diff --git a/builtin/repack.c b/builtin/repack.c
index d1449cfe13..5f382aaf19 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -108,11 +108,12 @@ static int repack_config(const char *var, const char *value,
 }
 
 static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
+				   const struct write_pack_opts *opts,
 				   struct child_process *cmd,
-				   struct string_list *names,
-				   int local)
+				   struct string_list *names)
 {
 	FILE *out;
+	bool local = write_pack_opts_is_local(opts);
 	struct strbuf line = STRBUF_INIT;
 
 	out = xfdopen(cmd->out, "r");
@@ -128,7 +129,8 @@ static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
 		 */
 		if (local) {
 			item = string_list_append(names, line.buf);
-			item->util = generated_pack_populate(line.buf, packtmp);
+			item->util = generated_pack_populate(line.buf,
+							     opts->packtmp);
 		}
 	}
 	fclose(out);
@@ -147,7 +149,6 @@ static int write_filtered_pack(const struct write_pack_opts *opts,
 	FILE *in;
 	int ret;
 	const char *caret;
-	bool local = write_pack_opts_is_local(opts);
 	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
@@ -183,8 +184,8 @@ static int write_filtered_pack(const struct write_pack_opts *opts,
 		fprintf(in, "%s%s.pack\n", caret, item->string);
 	fclose(in);
 
-	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
-				       local);
+	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
+				       names);
 }
 
 static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
@@ -231,7 +232,6 @@ static int write_cruft_pack(const struct write_pack_opts *opts,
 	struct string_list_item *item;
 	FILE *in;
 	int ret;
-	bool local = write_pack_opts_is_local(opts);
 	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
 
 	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
@@ -279,8 +279,8 @@ static int write_cruft_pack(const struct write_pack_opts *opts,
 		fprintf(in, "%s.pack\n", item->string);
 	fclose(in);
 
-	return finish_pack_objects_cmd(existing->repo->hash_algo, &cmd, names,
-				       local);
+	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
+				       names);
 }
 
 int cmd_repack(int argc,
@@ -560,9 +560,17 @@ int cmd_repack(int argc,
 		fclose(in);
 	}
 
-	ret = finish_pack_objects_cmd(repo->hash_algo, &cmd, &names, 1);
-	if (ret)
-		goto cleanup;
+	{
+		struct write_pack_opts opts = {
+			.packdir = packdir,
+			.destination = packdir,
+			.packtmp = packtmp,
+		};
+		ret = finish_pack_objects_cmd(repo->hash_algo, &opts, &cmd,
+					      &names);
+		if (ret)
+			goto cleanup;
+	}
 
 	if (!names.nr) {
 		if (!po_args.quiet)
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 45/49] repack: move `finish_pack_objects_cmd()` out of the builtin

In a similar spirit as the previous commit(s), now that the function `finish_pack_objects_cmd()` has no explicit dependencies within the repack builtin, let's extract it.

This prepares us to extract the remaining two functions within the repack builtin that explicitly write packfiles, which are `write_cruft_pack()` and `write_filtered_pack()`, which will be done in the future commits.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 33 ---------------------------------
 repack.c         | 33 +++++++++++++++++++++++++++++++++
 repack.h         |  5 +++++
 3 files changed, 38 insertions(+), 33 deletions(-)
Show changes to 3 files +38 −33

builtin/repack.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 5f382aaf19..71abcfa0b7 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -107,39 +107,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-static int finish_pack_objects_cmd(const struct git_hash_algo *algop,
-				   const struct write_pack_opts *opts,
-				   struct child_process *cmd,
-				   struct string_list *names)
-{
-	FILE *out;
-	bool local = write_pack_opts_is_local(opts);
-	struct strbuf line = STRBUF_INIT;
-
-	out = xfdopen(cmd->out, "r");
-	while (strbuf_getline_lf(&line, out) != EOF) {
-		struct string_list_item *item;
-
-		if (line.len != algop->hexsz)
-			die(_("repack: Expecting full hex object ID lines only "
-			      "from pack-objects."));
-		/*
-		 * Avoid putting packs written outside of the repository in the
-		 * list of names.
-		 */
-		if (local) {
-			item = string_list_append(names, line.buf);
-			item->util = generated_pack_populate(line.buf,
-							     opts->packtmp);
-		}
-	}
-	fclose(out);
-
-	strbuf_release(&line);
-
-	return finish_command(cmd);
-}
-
 static int write_filtered_pack(const struct write_pack_opts *opts,
 			       struct existing_packs *existing,
 			       struct string_list *names)
diff --git a/repack.c b/repack.c
index d2ee9f2460..2c478970f3 100644
--- a/repack.c
+++ b/repack.c
@@ -82,6 +82,39 @@ bool write_pack_opts_is_local(const struct write_pack_opts *opts)
 	return starts_with(opts->destination, opts->packdir);
 }
 
+int finish_pack_objects_cmd(const struct git_hash_algo *algop,
+			    const struct write_pack_opts *opts,
+			    struct child_process *cmd,
+			    struct string_list *names)
+{
+	FILE *out;
+	bool local = write_pack_opts_is_local(opts);
+	struct strbuf line = STRBUF_INIT;
+
+	out = xfdopen(cmd->out, "r");
+	while (strbuf_getline_lf(&line, out) != EOF) {
+		struct string_list_item *item;
+
+		if (line.len != algop->hexsz)
+			die(_("repack: Expecting full hex object ID lines only "
+			      "from pack-objects."));
+		/*
+		 * Avoid putting packs written outside of the repository in the
+		 * list of names.
+		 */
+		if (local) {
+			item = string_list_append(names, line.buf);
+			item->util = generated_pack_populate(line.buf,
+							     opts->packtmp);
+		}
+	}
+	fclose(out);
+
+	strbuf_release(&line);
+
+	return finish_command(cmd);
+}
+
 #define DELETE_PACK 1
 #define RETAIN_PACK 2
 
diff --git a/repack.h b/repack.h
index 26d1954ae2..3244f601e2 100644
--- a/repack.h
+++ b/repack.h
@@ -42,6 +42,11 @@ struct write_pack_opts {
 const char *write_pack_opts_pack_prefix(const struct write_pack_opts *opts);
 bool write_pack_opts_is_local(const struct write_pack_opts *opts);
 
+int finish_pack_objects_cmd(const struct git_hash_algo *algop,
+			    const struct write_pack_opts *opts,
+			    struct child_process *cmd,
+			    struct string_list *names);
+
 struct repository;
 struct packed_git;
 
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 46/49] repack: move `pack_kept_objects` to `struct pack_objects_args`

The "pack_kept_objects" variable is defined as static to the repack builtin, but is inherently related to the pack-objects arguments that the builtin uses when generating new packs.

Move that field into the "struct pack_objects_args", and shuffle around where we append the corresponding command-line option when preparing a pack-objects process. Specifically:

 - `write_cruft_pack()` always wants to pass "--honor-pack-keep", so
   explicitly set the `pack_kept_objects` field to "0" when initializing
   the `write_pack_opts` struct before calling `write_cruft_pack()`.
 - `write_filtered_pack()` no longer needs to handle writing the
   command-line option "--honor-pack-keep" when preparing a pack-objects
   process, since its call to `prepare_pack_objects()` will have already
   taken care of that.
   `write_filtered_pack()` also reads the `pack_kept_objects` field to
   determine whether to write the existing kept packs with a leading "^"
   character, so update that to read through the `po_args` pointer
   instead.
 - `cmd_repack()` also no longer has to write the "--honor-pack-keep"
   flag explicitly, since this is also handled via its call to
   `prepare_pack_objects()`.

Since there is a default value for "pack_kept_objects" that relies on whether or not we are writing a bitmap (and not writing a MIDX), extract a default initializer for `struct pack_objects_args` that keeps this conditional default behavior.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c  | 20 +++++++-------------
 repack-geometry.c |  5 ++---
 repack.c          |  2 ++
 repack.h          |  9 ++++++---
 4 files changed, 17 insertions(+), 19 deletions(-)
Show changes to 4 files +17 −19

builtin/repack.c, repack-geometry.c, repack.c, repack.h

diff --git a/builtin/repack.c b/builtin/repack.c
index 71abcfa0b7..3c6d7e91fd 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -33,7 +33,6 @@
 #define RETAIN_PACK 2
 
 static int pack_everything;
-static int pack_kept_objects = -1;
 static int write_bitmaps = -1;
 static int use_delta_islands;
 static int run_update_server_info = 1;
@@ -68,7 +67,7 @@ static int repack_config(const char *var, const char *value,
 		return 0;
 	}
 	if (!strcmp(var, "repack.packkeptobjects")) {
-		pack_kept_objects = git_config_bool(var, value);
+		po_args->pack_kept_objects = git_config_bool(var, value);
 		return 0;
 	}
 	if (!strcmp(var, "repack.writebitmaps") ||
@@ -122,8 +121,6 @@ static int write_filtered_pack(const struct write_pack_opts *opts,
 
 	strvec_push(&cmd.args, "--stdin-packs");
 
-	if (!pack_kept_objects)
-		strvec_push(&cmd.args, "--honor-pack-keep");
 	for_each_string_list_item(item, &existing->kept_packs)
 		strvec_pushf(&cmd.args, "--keep-pack=%s", item->string);
 
@@ -146,7 +143,7 @@ static int write_filtered_pack(const struct write_pack_opts *opts,
 		fprintf(in, "%s.pack\n", item->string);
 	for_each_string_list_item(item, &existing->cruft_packs)
 		fprintf(in, "%s.pack\n", item->string);
-	caret = pack_kept_objects ? "" : "^";
+	caret = opts->po_args->pack_kept_objects ? "" : "^";
 	for_each_string_list_item(item, &existing->kept_packs)
 		fprintf(in, "%s%s.pack\n", caret, item->string);
 	fclose(in);
@@ -208,7 +205,6 @@ static int write_cruft_pack(const struct write_pack_opts *opts,
 		strvec_pushf(&cmd.args, "--cruft-expiration=%s",
 			     cruft_expiration);
 
-	strvec_push(&cmd.args, "--honor-pack-keep");
 	strvec_push(&cmd.args, "--non-empty");
 
 	cmd.in = -1;
@@ -332,7 +328,7 @@ int cmd_repack(int argc,
 		OPT_UNSIGNED(0, "max-pack-size", &po_args.max_pack_size,
 			     N_("maximum size of each packfile")),
 		OPT_PARSE_LIST_OBJECTS_FILTER(&po_args.filter_options),
-		OPT_BOOL(0, "pack-kept-objects", &pack_kept_objects,
+		OPT_BOOL(0, "pack-kept-objects", &po_args.pack_kept_objects,
 				N_("repack objects in packs marked with .keep")),
 		OPT_STRING_LIST(0, "keep-pack", &keep_pack_list, N_("name"),
 				N_("do not repack this pack")),
@@ -378,8 +374,8 @@ int cmd_repack(int argc,
 		    (!(pack_everything & ALL_INTO_ONE) || !is_bare_repository()))
 			write_bitmaps = 0;
 	}
-	if (pack_kept_objects < 0)
-		pack_kept_objects = write_bitmaps > 0 && !write_midx;
+	if (po_args.pack_kept_objects < 0)
+		po_args.pack_kept_objects = write_bitmaps > 0 && !write_midx;
 
 	if (write_bitmaps && !(pack_everything & ALL_INTO_ONE) && !write_midx)
 		die(_(incremental_bitmap_conflict_error));
@@ -420,8 +416,7 @@ int cmd_repack(int argc,
 	if (geometry.split_factor) {
 		if (pack_everything)
 			die(_("options '%s' and '%s' cannot be used together"), "--geometric", "-A/-a");
-		pack_geometry_init(&geometry, &existing, &po_args,
-				   pack_kept_objects);
+		pack_geometry_init(&geometry, &existing, &po_args);
 		pack_geometry_split(&geometry);
 	}
 
@@ -430,8 +425,6 @@ int cmd_repack(int argc,
 	show_progress = !po_args.quiet && isatty(2);
 
 	strvec_push(&cmd.args, "--keep-true-parents");
-	if (!pack_kept_objects)
-		strvec_push(&cmd.args, "--honor-pack-keep");
 	for (i = 0; i < keep_pack_list.nr; i++)
 		strvec_pushf(&cmd.args, "--keep-pack=%s",
 			     keep_pack_list.items[i].string);
@@ -581,6 +574,7 @@ int cmd_repack(int argc,
 		cruft_po_args.local = po_args.local;
 		cruft_po_args.quiet = po_args.quiet;
 		cruft_po_args.delta_base_offset = po_args.delta_base_offset;
+		cruft_po_args.pack_kept_objects = 0;
 
 		ret = write_cruft_pack(&opts, cruft_expiration,
 				       combine_cruft_below_size, &names,
diff --git a/repack-geometry.c b/repack-geometry.c
index f58f1fc7f0..e2f9794d7d 100644
--- a/repack-geometry.c
+++ b/repack-geometry.c
@@ -27,8 +27,7 @@ static int pack_geometry_cmp(const void *va, const void *vb)
 
 void pack_geometry_init(struct pack_geometry *geometry,
 			struct existing_packs *existing,
-			const struct pack_objects_args *args,
-			int pack_kept_objects)
+			const struct pack_objects_args *args)
 {
 	struct packfile_store *packs = existing->repo->objects->packfiles;
 	struct packed_git *p;
@@ -43,7 +42,7 @@ void pack_geometry_init(struct pack_geometry *geometry,
 			 */
 			continue;
 
-		if (!pack_kept_objects) {
+		if (!args->pack_kept_objects) {
 			/*
 			 * Any pack that has its pack_keep bit set will
 			 * appear in existing->kept_packs below, but
diff --git a/repack.c b/repack.c
index 2c478970f3..2ab33c665a 100644
--- a/repack.c
+++ b/repack.c
@@ -38,6 +38,8 @@ void prepare_pack_objects(struct child_process *cmd,
 		strvec_push(&cmd->args,  "--quiet");
 	if (args->delta_base_offset)
 		strvec_push(&cmd->args,  "--delta-base-offset");
+	if (!args->pack_kept_objects)
+		strvec_push(&cmd->args,  "--honor-pack-keep");
 	strvec_push(&cmd->args, out);
 	cmd->git_cmd = 1;
 	cmd->out = -1;
diff --git a/repack.h b/repack.h
index 3244f601e2..0432379815 100644
--- a/repack.h
+++ b/repack.h
@@ -17,10 +17,14 @@ struct pack_objects_args {
 	int name_hash_version;
 	int path_walk;
 	int delta_base_offset;
+	int pack_kept_objects;
 	struct list_objects_filter_options filter_options;
 };
 
-#define PACK_OBJECTS_ARGS_INIT { .delta_base_offset = 1 }
+#define PACK_OBJECTS_ARGS_INIT { \
+	.delta_base_offset = 1, \
+	.pack_kept_objects = -1, \
+}
 
 struct child_process;
 
@@ -104,8 +108,7 @@ struct pack_geometry {
 
 void pack_geometry_init(struct pack_geometry *geometry,
 			struct existing_packs *existing,
-			const struct pack_objects_args *args,
-			int pack_kept_objects);
+			const struct pack_objects_args *args);
 void pack_geometry_split(struct pack_geometry *geometry);
 struct packed_git *pack_geometry_preferred_pack(struct pack_geometry *geometry);
 void pack_geometry_remove_redundant(struct pack_geometry *geometry,
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 47/49] repack: move `write_filtered_pack()` out of the builtin

In a similar fashion as in previous commits, move the function `write_filtered_pack()` out of the builtin and into its own compilation unit.

This function is now part of the repack.h API, but implemented in its own "repack-filtered.c" unit as it is a separate component from other kinds of repacking operations.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile          |  1 +
 builtin/repack.c  | 46 ------------------------------------------
 meson.build       |  1 +
 repack-filtered.c | 51 +++++++++++++++++++++++++++++++++++++++++++++++
 repack.h          |  4 ++++
 5 files changed, 57 insertions(+), 46 deletions(-)
 create mode 100644 repack-filtered.c
Show changes to 5 files +57 −46

Makefile, builtin/repack.c, meson.build, repack-filtered.c, repack.h

diff --git a/Makefile b/Makefile
index 4b0a122428..42461ea91f 100644
--- a/Makefile
+++ b/Makefile
@@ -1250,6 +1250,7 @@ LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
+LIB_OBJS += repack-filtered.o
 LIB_OBJS += repack-geometry.o
 LIB_OBJS += repack-midx.o
 LIB_OBJS += repack-promisor.o
diff --git a/builtin/repack.c b/builtin/repack.c
index 3c6d7e91fd..f65880d8f6 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -106,52 +106,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-static int write_filtered_pack(const struct write_pack_opts *opts,
-			       struct existing_packs *existing,
-			       struct string_list *names)
-{
-	struct child_process cmd = CHILD_PROCESS_INIT;
-	struct string_list_item *item;
-	FILE *in;
-	int ret;
-	const char *caret;
-	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
-
-	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
-
-	strvec_push(&cmd.args, "--stdin-packs");
-
-	for_each_string_list_item(item, &existing->kept_packs)
-		strvec_pushf(&cmd.args, "--keep-pack=%s", item->string);
-
-	cmd.in = -1;
-
-	ret = start_command(&cmd);
-	if (ret)
-		return ret;
-
-	/*
-	 * Here 'names' contains only the pack(s) that were just
-	 * written, which is exactly the packs we want to keep. Also
-	 * 'existing_kept_packs' already contains the packs in
-	 * 'keep_pack_list'.
-	 */
-	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, names)
-		fprintf(in, "^%s-%s.pack\n", pack_prefix, item->string);
-	for_each_string_list_item(item, &existing->non_kept_packs)
-		fprintf(in, "%s.pack\n", item->string);
-	for_each_string_list_item(item, &existing->cruft_packs)
-		fprintf(in, "%s.pack\n", item->string);
-	caret = opts->po_args->pack_kept_objects ? "" : "^";
-	for_each_string_list_item(item, &existing->kept_packs)
-		fprintf(in, "%s%s.pack\n", caret, item->string);
-	fclose(in);
-
-	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
-				       names);
-}
-
 static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
 				      struct existing_packs *existing)
 {
diff --git a/meson.build b/meson.build
index 0373d5c454..f5325b5941 100644
--- a/meson.build
+++ b/meson.build
@@ -463,6 +463,7 @@ libgit_sources = [
   'reftable/writer.c',
   'remote.c',
   'repack.c',
+  'repack-filtered.c',
   'repack-geometry.c',
   'repack-midx.c',
   'repack-promisor.c',
diff --git a/repack-filtered.c b/repack-filtered.c
new file mode 100644
index 0000000000..edcf7667c5
--- /dev/null
+++ b/repack-filtered.c
@@ -0,0 +1,51 @@
+#include "git-compat-util.h"
+#include "repack.h"
+#include "repository.h"
+#include "run-command.h"
+#include "string-list.h"
+
+int write_filtered_pack(const struct write_pack_opts *opts,
+			struct existing_packs *existing,
+			struct string_list *names)
+{
+	struct child_process cmd = CHILD_PROCESS_INIT;
+	struct string_list_item *item;
+	FILE *in;
+	int ret;
+	const char *caret;
+	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
+
+	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
+
+	strvec_push(&cmd.args, "--stdin-packs");
+
+	for_each_string_list_item(item, &existing->kept_packs)
+		strvec_pushf(&cmd.args, "--keep-pack=%s", item->string);
+
+	cmd.in = -1;
+
+	ret = start_command(&cmd);
+	if (ret)
+		return ret;
+
+	/*
+	 * Here 'names' contains only the pack(s) that were just
+	 * written, which is exactly the packs we want to keep. Also
+	 * 'existing_kept_packs' already contains the packs in
+	 * 'keep_pack_list'.
+	 */
+	in = xfdopen(cmd.in, "w");
+	for_each_string_list_item(item, names)
+		fprintf(in, "^%s-%s.pack\n", pack_prefix, item->string);
+	for_each_string_list_item(item, &existing->non_kept_packs)
+		fprintf(in, "%s.pack\n", item->string);
+	for_each_string_list_item(item, &existing->cruft_packs)
+		fprintf(in, "%s.pack\n", item->string);
+	caret = opts->po_args->pack_kept_objects ? "" : "^";
+	for_each_string_list_item(item, &existing->kept_packs)
+		fprintf(in, "%s%s.pack\n", caret, item->string);
+	fclose(in);
+
+	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
+				       names);
+}
diff --git a/repack.h b/repack.h
index 0432379815..c790c90ef8 100644
--- a/repack.h
+++ b/repack.h
@@ -133,4 +133,8 @@ struct repack_write_midx_opts {
 void midx_snapshot_refs(struct repository *repo, struct tempfile *f);
 int write_midx_included_packs(struct repack_write_midx_opts *opts);
 
+int write_filtered_pack(const struct write_pack_opts *opts,
+			struct existing_packs *existing,
+			struct string_list *names);
+
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 48/49] repack: move `write_cruft_pack()` out of the builtin

In an identical fashion as the previous commit, move the function `write_cruft_pack()` into its own compilation unit, and make the function visible through the repack.h API.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 Makefile         |  1 +
 builtin/repack.c | 94 ---------------------------------------------
 meson.build      |  1 +
 repack-cruft.c   | 99 ++++++++++++++++++++++++++++++++++++++++++++++++
 repack.h         |  6 +++
 5 files changed, 107 insertions(+), 94 deletions(-)
 create mode 100644 repack-cruft.c
Show changes to 5 files +107 −94

Makefile, builtin/repack.c, meson.build, repack-cruft.c, repack.h

diff --git a/Makefile b/Makefile
index 42461ea91f..fb965e6acb 100644
--- a/Makefile
+++ b/Makefile
@@ -1250,6 +1250,7 @@ LIB_OBJS += refs/ref-cache.o
 LIB_OBJS += refspec.o
 LIB_OBJS += remote.o
 LIB_OBJS += repack.o
+LIB_OBJS += repack-cruft.o
 LIB_OBJS += repack-filtered.o
 LIB_OBJS += repack-geometry.o
 LIB_OBJS += repack-midx.o
diff --git a/builtin/repack.c b/builtin/repack.c
index f65880d8f6..a68c22f605 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -106,100 +106,6 @@ static int repack_config(const char *var, const char *value,
 	return git_default_config(var, value, ctx, cb);
 }
 
-static void combine_small_cruft_packs(FILE *in, size_t combine_cruft_below_size,
-				      struct existing_packs *existing)
-{
-	struct packfile_store *packs = existing->repo->objects->packfiles;
-	struct packed_git *p;
-	struct strbuf buf = STRBUF_INIT;
-	size_t i;
-
-	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
-		if (!(p->is_cruft && p->pack_local))
-			continue;
-
-		strbuf_reset(&buf);
-		strbuf_addstr(&buf, pack_basename(p));
-		strbuf_strip_suffix(&buf, ".pack");
-
-		if (!string_list_has_string(&existing->cruft_packs, buf.buf))
-			continue;
-
-		if (p->pack_size < combine_cruft_below_size) {
-			fprintf(in, "-%s\n", pack_basename(p));
-		} else {
-			existing_packs_retain_cruft(existing, p);
-			fprintf(in, "%s\n", pack_basename(p));
-		}
-	}
-
-	for (i = 0; i < existing->non_kept_packs.nr; i++)
-		fprintf(in, "-%s.pack\n",
-			existing->non_kept_packs.items[i].string);
-
-	strbuf_release(&buf);
-}
-
-static int write_cruft_pack(const struct write_pack_opts *opts,
-			    const char *cruft_expiration,
-			    unsigned long combine_cruft_below_size,
-			    struct string_list *names,
-			    struct existing_packs *existing)
-{
-	struct child_process cmd = CHILD_PROCESS_INIT;
-	struct string_list_item *item;
-	FILE *in;
-	int ret;
-	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
-
-	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
-
-	strvec_push(&cmd.args, "--cruft");
-	if (cruft_expiration)
-		strvec_pushf(&cmd.args, "--cruft-expiration=%s",
-			     cruft_expiration);
-
-	strvec_push(&cmd.args, "--non-empty");
-
-	cmd.in = -1;
-
-	ret = start_command(&cmd);
-	if (ret)
-		return ret;
-
-	/*
-	 * names has a confusing double use: it both provides the list
-	 * of just-written new packs, and accepts the name of the cruft
-	 * pack we are writing.
-	 *
-	 * By the time it is read here, it contains only the pack(s)
-	 * that were just written, which is exactly the set of packs we
-	 * want to consider kept.
-	 *
-	 * If `--expire-to` is given, the double-use served by `names`
-	 * ensures that the pack written to `--expire-to` excludes any
-	 * objects contained in the cruft pack.
-	 */
-	in = xfdopen(cmd.in, "w");
-	for_each_string_list_item(item, names)
-		fprintf(in, "%s-%s.pack\n", pack_prefix, item->string);
-	if (combine_cruft_below_size && !cruft_expiration) {
-		combine_small_cruft_packs(in, combine_cruft_below_size,
-					  existing);
-	} else {
-		for_each_string_list_item(item, &existing->non_kept_packs)
-			fprintf(in, "-%s.pack\n", item->string);
-		for_each_string_list_item(item, &existing->cruft_packs)
-			fprintf(in, "-%s.pack\n", item->string);
-	}
-	for_each_string_list_item(item, &existing->kept_packs)
-		fprintf(in, "%s.pack\n", item->string);
-	fclose(in);
-
-	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
-				       names);
-}
-
 int cmd_repack(int argc,
 	       const char **argv,
 	       const char *prefix,
diff --git a/meson.build b/meson.build
index f5325b5941..81a3d5b96a 100644
--- a/meson.build
+++ b/meson.build
@@ -463,6 +463,7 @@ libgit_sources = [
   'reftable/writer.c',
   'remote.c',
   'repack.c',
+  'repack-cruft.c',
   'repack-filtered.c',
   'repack-geometry.c',
   'repack-midx.c',
diff --git a/repack-cruft.c b/repack-cruft.c
new file mode 100644
index 0000000000..c51df36722
--- /dev/null
+++ b/repack-cruft.c
@@ -0,0 +1,99 @@
+#include "git-compat-util.h"
+#include "repack.h"
+#include "packfile.h"
+#include "repository.h"
+#include "run-command.h"
+
+static void combine_small_cruft_packs(FILE *in, off_t combine_cruft_below_size,
+				      struct existing_packs *existing)
+{
+	struct packfile_store *packs = existing->repo->objects->packfiles;
+	struct packed_git *p;
+	struct strbuf buf = STRBUF_INIT;
+	size_t i;
+
+	for (p = packfile_store_get_all_packs(packs); p; p = p->next) {
+		if (!(p->is_cruft && p->pack_local))
+			continue;
+
+		strbuf_reset(&buf);
+		strbuf_addstr(&buf, pack_basename(p));
+		strbuf_strip_suffix(&buf, ".pack");
+
+		if (!string_list_has_string(&existing->cruft_packs, buf.buf))
+			continue;
+
+		if (p->pack_size < combine_cruft_below_size) {
+			fprintf(in, "-%s\n", pack_basename(p));
+		} else {
+			existing_packs_retain_cruft(existing, p);
+			fprintf(in, "%s\n", pack_basename(p));
+		}
+	}
+
+	for (i = 0; i < existing->non_kept_packs.nr; i++)
+		fprintf(in, "-%s.pack\n",
+			existing->non_kept_packs.items[i].string);
+
+	strbuf_release(&buf);
+}
+
+int write_cruft_pack(const struct write_pack_opts *opts,
+		     const char *cruft_expiration,
+		     unsigned long combine_cruft_below_size,
+		     struct string_list *names,
+		     struct existing_packs *existing)
+{
+	struct child_process cmd = CHILD_PROCESS_INIT;
+	struct string_list_item *item;
+	FILE *in;
+	int ret;
+	const char *pack_prefix = write_pack_opts_pack_prefix(opts);
+
+	prepare_pack_objects(&cmd, opts->po_args, opts->destination);
+
+	strvec_push(&cmd.args, "--cruft");
+	if (cruft_expiration)
+		strvec_pushf(&cmd.args, "--cruft-expiration=%s",
+			     cruft_expiration);
+
+	strvec_push(&cmd.args, "--non-empty");
+
+	cmd.in = -1;
+
+	ret = start_command(&cmd);
+	if (ret)
+		return ret;
+
+	/*
+	 * names has a confusing double use: it both provides the list
+	 * of just-written new packs, and accepts the name of the cruft
+	 * pack we are writing.
+	 *
+	 * By the time it is read here, it contains only the pack(s)
+	 * that were just written, which is exactly the set of packs we
+	 * want to consider kept.
+	 *
+	 * If `--expire-to` is given, the double-use served by `names`
+	 * ensures that the pack written to `--expire-to` excludes any
+	 * objects contained in the cruft pack.
+	 */
+	in = xfdopen(cmd.in, "w");
+	for_each_string_list_item(item, names)
+		fprintf(in, "%s-%s.pack\n", pack_prefix, item->string);
+	if (combine_cruft_below_size && !cruft_expiration) {
+		combine_small_cruft_packs(in, combine_cruft_below_size,
+					  existing);
+	} else {
+		for_each_string_list_item(item, &existing->non_kept_packs)
+			fprintf(in, "-%s.pack\n", item->string);
+		for_each_string_list_item(item, &existing->cruft_packs)
+			fprintf(in, "-%s.pack\n", item->string);
+	}
+	for_each_string_list_item(item, &existing->kept_packs)
+		fprintf(in, "%s.pack\n", item->string);
+	fclose(in);
+
+	return finish_pack_objects_cmd(existing->repo->hash_algo, opts, &cmd,
+				       names);
+}
diff --git a/repack.h b/repack.h
index c790c90ef8..3a688a12ee 100644
--- a/repack.h
+++ b/repack.h
@@ -137,4 +137,10 @@ int write_filtered_pack(const struct write_pack_opts *opts,
 			struct existing_packs *existing,
 			struct string_list *names);
 
+int write_cruft_pack(const struct write_pack_opts *opts,
+		     const char *cruft_expiration,
+		     unsigned long combine_cruft_below_size,
+		     struct string_list *names,
+		     struct existing_packs *existing);
+
 #endif /* REPACK_H */
-- 
2.51.0.540.ga7423965ad8
Taylor Blau· Oct 15, 2025, 22:29 UTC · re: Taylor Blau · lore

[PATCH v2 49/49] builtin/repack.c: clean up unused `#include`s

Over the past several dozen commits, we have moved a large amount of functionality out of the repack builtin and into other files like repack.c, repack-cruft.c, repack-filtered.c, repack-midx.c, and repack-promisor.c.

These files specify the minimal set of `#include`s that they need to compile successfully, but we did not change the set of `#include`s in the repack builtin itself.

Now that the code movement is complete, let's clean up that set of `#include`s and trim down the builtin to include the minimal amount of external headers necessary to compile.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
---
 builtin/repack.c | 9 ---------
 1 file changed, 9 deletions(-)
Show changes to builtin/repack.c +0 −9
diff --git a/builtin/repack.c b/builtin/repack.c
index a68c22f605..cfdb4c0920 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -4,26 +4,17 @@
 #include "builtin.h"
 #include "config.h"
 #include "environment.h"
-#include "gettext.h"
-#include "hex.h"
 #include "parse-options.h"
 #include "path.h"
 #include "run-command.h"
 #include "server-info.h"
-#include "strbuf.h"
 #include "string-list.h"
-#include "strvec.h"
 #include "midx.h"
 #include "packfile.h"
 #include "prune-packed.h"
-#include "odb.h"
 #include "promisor-remote.h"
 #include "repack.h"
 #include "shallow.h"
-#include "pack.h"
-#include "pack-bitmap.h"
-#include "refs.h"
-#include "list-objects-filter-options.h"
 
 #define ALL_INTO_ONE 1
 #define LOOSEN_UNREACHABLE 2
-- 
2.51.0.540.ga7423965ad8
Patrick Steinhardt· Oct 16, 2025, 10:31 UTC · re: Taylor Blau · lore

Re: [PATCH v2 00/49] repack: prepare for incremental MIDX-based repacking

On Wed, Oct 15, 2025 at 06:26:57PM -0400, Taylor Blau wrote:
Show 32 quoted lines
> Note to the maintainer:
> 
>  * This series has been rebased on to 'master' (which is 4b71b29477
>    (The seventeenth batch, 2025-10-10) at the time of writing) to
>    avoid semantic conflicts now that ps/packfile-store has been
>    merged.
> 
> This is a relatively small reroll of my series to clean up the repack
> builtin by introducing a repack.h API, which is the first of ~three
> series that will implement incremental MIDX/bitmap-based repacking.
> 
> A range-diff is included below for convenience, but the changes since
> v1 are generally limited to the following:
> 
>  * Wording tweaks and a couple of minor typo fixes.
> 
>  * Dropping explicit casts out of 'void *'.
> 
>  * Clarification in commit "builtin/repack.c: introduce `struct
>    write_pack_opts`" that additional cleanup follows in the upcoming
>    patches.
> 
>  * Marking parts of the new API as const where possible.
> 
>  * Using 'bool' as the return type and simplifying the implementation
>    of `write_pack_opts_is_local()`.
> 
>  * Avoid shadowing "struct write_pack_opts opts" in `cmd_repack()`.
> 
> Outside of that, the series is unchanged, and I am hopeful that this
> round looks good to reviewers so that we can move on to the more
> interesting parts of incremental MIDX/bitmap repacking ;-).

I didn't quite feel like reviewing all of these patches again, so I only had a look at the range-diff. The changes in there all look good to me and address my feedback.

So I feel like all of this is in a reasonably good shape and a definitive improvement for our code base. Thanks!

Patrick
Junio C Hamano· Oct 16, 2025, 17:11 UTC · re: Patrick Steinhardt · lore

Re: [PATCH v2 00/49] repack: prepare for incremental MIDX-based repacking

Patrick Steinhardt <ps@pks.im> writes:
Show 28 quoted lines
>> A range-diff is included below for convenience, but the changes since
>> v1 are generally limited to the following:
>> 
>>  * Wording tweaks and a couple of minor typo fixes.
>> 
>>  * Dropping explicit casts out of 'void *'.
>> 
>>  * Clarification in commit "builtin/repack.c: introduce `struct
>>    write_pack_opts`" that additional cleanup follows in the upcoming
>>    patches.
>> 
>>  * Marking parts of the new API as const where possible.
>> 
>>  * Using 'bool' as the return type and simplifying the implementation
>>    of `write_pack_opts_is_local()`.
>> 
>>  * Avoid shadowing "struct write_pack_opts opts" in `cmd_repack()`.
>> 
>> Outside of that, the series is unchanged, and I am hopeful that this
>> round looks good to reviewers so that we can move on to the more
>> interesting parts of incremental MIDX/bitmap repacking ;-).
>
> I didn't quite feel like reviewing all of these patches again, so I only
> had a look at the range-diff. The changes in there all look good to me
> and address my feedback.
>
> So I feel like all of this is in a reasonably good shape and a
> definitive improvement for our code base. Thanks!
;-)

Thanks for a vote of confidence. With what I saw in v1 round, I agree these are in good shape.

Thanks.
Taylor Blau· Oct 16, 2025, 19:56 UTC · re: Junio C Hamano · lore

Re: [PATCH v2 00/49] repack: prepare for incremental MIDX-based repacking

On Thu, Oct 16, 2025 at 10:11:19AM -0700, Junio C Hamano wrote:
Show 11 quoted lines
> > I didn't quite feel like reviewing all of these patches again, so I only
> > had a look at the range-diff. The changes in there all look good to me
> > and address my feedback.
> >
> > So I feel like all of this is in a reasonably good shape and a
> > definitive improvement for our code base. Thanks!
>
> ;-)
>
> Thanks for a vote of confidence.  With what I saw in v1 round, I
> agree these are in good shape.
Thanks, both :-).

Thanks, Taylor

Jeff King· Oct 17, 2025, 08:40 UTC · re: Junio C Hamano · lore

Re: [PATCH v2 00/49] repack: prepare for incremental MIDX-based repacking

On Thu, Oct 16, 2025 at 10:11:19AM -0700, Junio C Hamano wrote:
Show 11 quoted lines
> > I didn't quite feel like reviewing all of these patches again, so I only
> > had a look at the range-diff. The changes in there all look good to me
> > and address my feedback.
> >
> > So I feel like all of this is in a reasonably good shape and a
> > definitive improvement for our code base. Thanks!
> 
> ;-)
> 
> Thanks for a vote of confidence.  With what I saw in v1 round, I
> agree these are in good shape.

I likewise looked over the range-diff, and it covered all of my actionable comments from the v1 review.

-Peff

← back to recent threads