{"thread":{"id":"57949","subject":"[PATCH 0/4] rebase: update branches in multi-part topic","startedAt":"2022-06-03T13:37:59Z","lastAt":"2022-07-21T20:05:06Z","messageCount":144,"participants":["Derrick Stolee via GitGitGadget","Junio C Hamano","Derrick Stolee","Taylor Blau","Jeff Hostetler","Phillip Wood","Elijah Newren","SZEDER Gábor"],"isPatch":true,"patchVersion":1,"patchTotal":4},"messages":[{"id":"456576","messageId":"pull.1247.git.1654263472.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":null,"subject":"[PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-03T13:37:48Z","receivedAt":"2022-06-03T13:37:59Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"This is a feature I've wanted for quite a while. When working on the sparse\nindex topic, I created a long RFC that actually broke into three topics for\nfull review upstream. These topics were sequential, so any feedback on an\nearlier one required updates to the later ones. I would work on the full\nfeature and use interactive rebase to update the full list of commits.\nHowever, I would need to update the branches pointing to those sub-topics.\n\nThis series adds a new --update-refs option to 'git rebase' (along with a\nrebase.updateRefs config option) that adds 'git update-ref' commands into\nthe TODO list. This is powered by the commit decoration machinery.\n\nAs an example, here is my in-progress bundle URI RFC split into subtopics as\nthey appear during the TODO list of a git rebase -i --update-refs:\n\npick 2d966282ff3 docs: document bundle URI standard\npick 31396e9171a remote-curl: add 'get' capability\npick 54c6ab70f67 bundle-uri: create basic file-copy logic\npick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\npick 6adaf842684 fetch: add --bundle-uri option\npick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\nexec git update-ref refs/heads/bundle-redo/fetch HEAD 6c5840ed77e1bc41c1fe6fb7c894ceede1b8d730\n\npick 1e3f6546632 clone: add --bundle-uri option\npick 9e4a6fe9b68 clone: --bundle-uri cannot be combined with --depth\nexec git update-ref refs/heads/bundle-redo/clone HEAD 9e4a6fe9b68a8455b427c9ac8cdbff30c96653b4\n\npick 5451cb6599c bundle-uri: create bundle_list struct and helpers\npick 3029c3aca15 bundle-uri: create base key-value pair parsing\npick a8b2de79ce8 bundle-uri: create \"key=value\" line parsing\npick 92625a47673 bundle-uri: unit test \"key=value\" parsing\npick a8616af4dc2 bundle-uri: limit recursion depth for bundle lists\npick 9d6809a8d53 bundle-uri: parse bundle list in config format\npick 287a732b54c bundle-uri: fetch a list of bundles\nexec git update-ref refs/heads/bundle-redo/list HEAD 287a732b54c4d95e7f410b3b36ef90d8a19cd346\n\npick b09f8226185 protocol v2: add server-side \"bundle-uri\" skeleton\npick 520204dcd1c bundle-uri client: add minimal NOOP client\npick 62e8b457b48 bundle-uri client: add \"git ls-remote-bundle-uri\"\npick 00eae925043 bundle-uri: serve URI advertisement from bundle.* config\npick 4277440a250 bundle-uri client: add boolean transfer.bundleURI setting\npick caf4599a81d bundle-uri: allow relative URLs in bundle lists\npick df255000b7e bundle-uri: download bundles from an advertised list\npick d71beabf199 clone: unbundle the advertised bundles\npick c9578391976 t5601: basic bundle URI tests\n# Ref refs/heads/bundle-redo/rfc-3 checked out at '/home/stolee/_git/git-bundles'\n\nexec git update-ref refs/heads/bundle-redo/advertise HEAD c9578391976ab9899c4e4f9b5fa2827650097305\n\n\nThe first two patches are helpers that are needed, but the full logic of the\n--update-refs option is introduced in patch 3. The config option is\navailable in patch 4.\n\nThanks, -Stolee\n\nDerrick Stolee (4):\n log-tree: create for_each_decoration()\n branch: add branch_checked_out() helper\n rebase: add --update-refs option\n rebase: add rebase.updateRefs config option\n\n Documentation/config/rebase.txt | 3 +\n Documentation/git-rebase.txt | 11 ++++\n branch.c | 24 ++++---\n branch.h | 8 +++\n builtin/rebase.c | 10 +++\n log-tree.c | 111 ++++++++++++++++++++++----------\n log-tree.h | 4 ++\n sequencer.c | 99 ++++++++++++++++++++++++++++\n sequencer.h | 1 +\n t/t3404-rebase-interactive.sh | 34 ++++++++++\n 10 files changed, 262 insertions(+), 43 deletions(-)\n\n\nbase-commit: 2668e3608e47494f2f10ef2b6e69f08a84816bcb\nPublished-As: https://github.com/gitgitgadget/git/releases/tag/pr-1247%2Fderrickstolee%2Frebase-keep-decorations-v1\nFetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1247/derrickstolee/rebase-keep-decorations-v1\nPull-Request: https://github.com/gitgitgadget/git/pull/1247\n-- \ngitgitgadget\n"},{"id":"456577","messageId":"5f54766e1032ebf3a331516a6dd696b997bdfdd8.1654263472.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.git.1654263472.gitgitgadget@gmail.com","subject":"[PATCH 2/4] branch: add branch_checked_out() helper","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-03T13:37:50Z","receivedAt":"2022-06-03T13:38:07Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe validate_new_branchname() method contains a check to see if a branch\nis checked out in any non-bare worktree. This is intended to prevent a\nforce push that will mess up an existing checkout. This helper is not\nsuitable to performing just that check, because the method will die()\nwhen the branch is checked out instead of returning an error code.\n\nExtract branch_checked_out() and use it within\nvalidate_new_branchname(). Another caller will be added in a coming\nchange.\n\nSigned-off-by: Derrick Stolee \n---\n branch.c | 24 ++++++++++++++++--------\n branch.h | 8 ++++++++\n 2 files changed, 24 insertions(+), 8 deletions(-)\n\ndiff --git a/branch.c b/branch.c\nindex 2d6569b0c62..2e6419cdfa5 100644\n--- a/branch.c\n+++ b/branch.c\n@@ -369,6 +369,19 @@ int validate_branchname(const char *name, struct strbuf *ref)\n \treturn ref_exists(ref->buf);\n }\n \n+int branch_checked_out(const char *refname, char **path)\n+{\n+\tstruct worktree **worktrees = get_worktrees();\n+\tconst struct worktree *wt = find_shared_symref(worktrees, \"HEAD\", refname);\n+\tint result = wt && !wt->is_bare;\n+\n+\tif (result && path)\n+\t\t*path = xstrdup(wt->path);\n+\n+\tfree_worktrees(worktrees);\n+\treturn result;\n+}\n+\n /*\n * Check if a branch 'name' can be created as a new branch; die otherwise.\n * 'force' can be used when it is OK for the named branch already exists.\n@@ -377,9 +390,7 @@ int validate_branchname(const char *name, struct strbuf *ref)\n */\n int validate_new_branchname(const char *name, struct strbuf *ref, int force)\n {\n-\tstruct worktree **worktrees;\n-\tconst struct worktree *wt;\n-\n+\tchar *path;\n \tif (!validate_branchname(name, ref))\n \t\treturn 0;\n \n@@ -387,13 +398,10 @@ int validate_new_branchname(const char *name, struct strbuf *ref, int force)\n \t\tdie(_(\"a branch named '%s' already exists\"),\n \t\t ref->buf + strlen(\"refs/heads/\"));\n \n-\tworktrees = get_worktrees();\n-\twt = find_shared_symref(worktrees, \"HEAD\", ref->buf);\n-\tif (wt && !wt->is_bare)\n+\tif (branch_checked_out(ref->buf, &path))\n \t\tdie(_(\"cannot force update the branch '%s' \"\n \t\t \"checked out at '%s'\"),\n-\t\t ref->buf + strlen(\"refs/heads/\"), wt->path);\n-\tfree_worktrees(worktrees);\n+\t\t ref->buf + strlen(\"refs/heads/\"), path);\n \n \treturn 1;\n }\ndiff --git a/branch.h b/branch.h\nindex 560b6b96a8f..5ea93d217b1 100644\n--- a/branch.h\n+++ b/branch.h\n@@ -101,6 +101,14 @@ void create_branches_recursively(struct repository *r, const char *name,\n \t\t\t\t const char *tracking_name, int force,\n \t\t\t\t int reflog, int quiet, enum branch_track track,\n \t\t\t\t int dry_run);\n+\n+/*\n+ * Returns true if the branch at 'refname' is checked out at any\n+ * non-bare worktree. The path of the worktree is stored in the\n+ * given 'path', if provided.\n+ */\n+int branch_checked_out(const char *refname, char **path);\n+\n /*\n * Check if 'name' can be a valid name for a branch; die otherwise.\n * Return 1 if the named branch already exists; return 0 otherwise.\n-- \ngitgitgadget\n\n"},{"id":"456578","messageId":"4f9f34876413927d819313a70fcdefcad5b35689.1654263472.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.git.1654263472.gitgitgadget@gmail.com","subject":"[PATCH 1/4] log-tree: create for_each_decoration()","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-03T13:37:49Z","receivedAt":"2022-06-03T13:38:08Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nIt can be helpful to iterate through all the decorations on a commit\nwithout necessarily writing them to a stream. Implement\nfor_each_decoration() and reimplement format_decorations_extended() to\nuse that iterator.\n\nSigned-off-by: Derrick Stolee \n---\n log-tree.c | 111 ++++++++++++++++++++++++++++++++++++-----------------\n log-tree.h | 4 ++\n 2 files changed, 80 insertions(+), 35 deletions(-)\n\ndiff --git a/log-tree.c b/log-tree.c\nindex d0ac0a6327a..b15a7c9db22 100644\n--- a/log-tree.c\n+++ b/log-tree.c\n@@ -282,6 +282,54 @@ static void show_name(struct strbuf *sb, const struct name_decoration *decoratio\n \t\tstrbuf_addstr(sb, decoration->name);\n }\n \n+struct format_decorations_context {\n+\tstruct strbuf *sb;\n+\tint use_color;\n+\tconst char *prefix;\n+\tconst char *separator;\n+\tconst char *suffix;\n+\tconst char *color_commit;\n+\tconst char *color_reset;\n+\tconst struct name_decoration *current_and_HEAD;\n+};\n+\n+static int append_decoration(const struct name_decoration *d,\n+\t\t\t void *data)\n+{\n+\tstruct format_decorations_context *ctx = data;\n+\t/*\n+\t * When both current and HEAD are there, only\n+\t * show HEAD->current where HEAD would have\n+\t * appeared, skipping the entry for current.\n+\t */\n+\tif (d != ctx->current_and_HEAD) {\n+\t\tstrbuf_addstr(ctx->sb, ctx->color_commit);\n+\t\tstrbuf_addstr(ctx->sb, ctx->prefix);\n+\t\tstrbuf_addstr(ctx->sb, ctx->color_reset);\n+\t\tstrbuf_addstr(ctx->sb, decorate_get_color(ctx->use_color, d->type));\n+\t\tif (d->type == DECORATION_REF_TAG)\n+\t\t\tstrbuf_addstr(ctx->sb, \"tag: \");\n+\n+\t\tshow_name(ctx->sb, d);\n+\n+\t\tif (ctx->current_and_HEAD &&\n+\t\t d->type == DECORATION_REF_HEAD) {\n+\t\t\tstrbuf_addstr(ctx->sb, \" -> \");\n+\t\t\tstrbuf_addstr(ctx->sb, ctx->color_reset);\n+\t\t\tstrbuf_addstr(ctx->sb,\n+\t\t\t\t decorate_get_color(\n+\t\t\t\t\t\tctx->use_color,\n+\t\t\t\t\t\tctx->current_and_HEAD->type));\n+\t\t\tshow_name(ctx->sb, ctx->current_and_HEAD);\n+\t\t}\n+\t\tstrbuf_addstr(ctx->sb, ctx->color_reset);\n+\n+\t\tctx->prefix = ctx->separator;\n+\t}\n+\n+\treturn 0;\n+}\n+\n /*\n * The caller makes sure there is no funny color before calling.\n * format_decorations_extended makes sure the same after return.\n@@ -294,49 +342,42 @@ void format_decorations_extended(struct strbuf *sb,\n \t\t\tconst char *suffix)\n {\n \tconst struct name_decoration *decoration;\n-\tconst struct name_decoration *current_and_HEAD;\n-\tconst char *color_commit =\n-\t\tdiff_get_color(use_color, DIFF_COMMIT);\n-\tconst char *color_reset =\n-\t\tdecorate_get_color(use_color, DECORATION_NONE);\n+\tstruct format_decorations_context ctx = {\n+\t\t.sb = sb,\n+\t\t.use_color = use_color,\n+\t\t.prefix = prefix,\n+\t\t.separator = separator,\n+\t\t.suffix = suffix,\n+\t\t.color_commit =\tdiff_get_color(use_color, DIFF_COMMIT),\n+\t\t.color_reset = decorate_get_color(use_color, DECORATION_NONE),\n+\t};\n \n \tdecoration = get_name_decoration(&commit->object);\n \tif (!decoration)\n \t\treturn;\n \n-\tcurrent_and_HEAD = current_pointed_by_HEAD(decoration);\n-\twhile (decoration) {\n-\t\t/*\n-\t\t * When both current and HEAD are there, only\n-\t\t * show HEAD->current where HEAD would have\n-\t\t * appeared, skipping the entry for current.\n-\t\t */\n-\t\tif (decoration != current_and_HEAD) {\n-\t\t\tstrbuf_addstr(sb, color_commit);\n-\t\t\tstrbuf_addstr(sb, prefix);\n-\t\t\tstrbuf_addstr(sb, color_reset);\n-\t\t\tstrbuf_addstr(sb, decorate_get_color(use_color, decoration->type));\n-\t\t\tif (decoration->type == DECORATION_REF_TAG)\n-\t\t\t\tstrbuf_addstr(sb, \"tag: \");\n-\n-\t\t\tshow_name(sb, decoration);\n-\n-\t\t\tif (current_and_HEAD &&\n-\t\t\t decoration->type == DECORATION_REF_HEAD) {\n-\t\t\t\tstrbuf_addstr(sb, \" -> \");\n-\t\t\t\tstrbuf_addstr(sb, color_reset);\n-\t\t\t\tstrbuf_addstr(sb, decorate_get_color(use_color, current_and_HEAD->type));\n-\t\t\t\tshow_name(sb, current_and_HEAD);\n-\t\t\t}\n-\t\t\tstrbuf_addstr(sb, color_reset);\n+\tctx.current_and_HEAD = current_pointed_by_HEAD(decoration);\n \n-\t\t\tprefix = separator;\n-\t\t}\n+\tfor_each_decoration(commit, append_decoration, &ctx);\n+\n+\tstrbuf_addstr(sb, ctx.color_commit);\n+\tstrbuf_addstr(sb, ctx.suffix);\n+\tstrbuf_addstr(sb, ctx.color_reset);\n+}\n+\n+int for_each_decoration(const struct commit *c, decoration_fn fn, void *data)\n+{\n+\tconst struct name_decoration *decoration;\n+\n+\tdecoration = get_name_decoration(&c->object);\n+\twhile (decoration) {\n+\t\tint res;\n+\t\tif ((res = fn(decoration, data)))\n+\t\t\treturn res;\n \t\tdecoration = decoration->next;\n \t}\n-\tstrbuf_addstr(sb, color_commit);\n-\tstrbuf_addstr(sb, suffix);\n-\tstrbuf_addstr(sb, color_reset);\n+\n+\treturn 0;\n }\n \n void show_decorations(struct rev_info *opt, struct commit *commit)\ndiff --git a/log-tree.h b/log-tree.h\nindex e7e4641cf83..ea07da2625b 100644\n--- a/log-tree.h\n+++ b/log-tree.h\n@@ -35,4 +35,8 @@ void fmt_output_commit(struct strbuf *, struct commit *, struct rev_info *);\n void fmt_output_subject(struct strbuf *, const char *subject, struct rev_info *);\n void fmt_output_email_subject(struct strbuf *, struct rev_info *);\n \n+typedef int decoration_fn(const struct name_decoration *d,\n+\t\t\t void *data);\n+int for_each_decoration(const struct commit *c, decoration_fn fn, void *data);\n+\n #endif\n-- \ngitgitgadget\n\n"},{"id":"456579","messageId":"23fa663886405596370e5fc95f679ef299f0725c.1654263472.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.git.1654263472.gitgitgadget@gmail.com","subject":"[PATCH 3/4] rebase: add --update-refs option","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-03T13:37:51Z","receivedAt":"2022-06-03T13:38:10Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nWhen working on a large feature, it can be helpful to break that feature\ninto multiple smaller parts that become reviewed in sequence. During\ndevelopment or during review, a change to one part of the feature could\naffect multiple of these parts. An interactive rebase can help adjust\nthe multi-part \"story\" of the branch.\n\nHowever, if there are branches tracking the different parts of the\nfeature, then rebasing the entire list of commits can create commits not\nreachable from those \"sub branches\". It can take a manual step to update\nthose branches.\n\nAdd a new --update-refs option to 'git rebase -i' that adds 'git\nupdate-ref' exec steps to the todo file whenever a commit that is being\nrebased is decorated with that . This allows the user to rebase a\nlong list of commits in a multi-part feature and keep all of their\npointers to those parts.\n\nUse the new for_each_decoration() while iterating over the existing todo\nlist. Be sure to iterate after any squashing or fixups are placed.\nUpdate the branch only after those squashes and fixups are complete.\nThis allows a --fixup commit at the tip of the feature to apply\ncorrectly to the sub branch, even if it is fixing up the most-recent\ncommit in that part.\n\nOne potential problem here is that refs decorating commits that are\nalready marked as \"fixup!\" or \"squash!\" will not be included in this\nlist. Generally, the reordering of the \"fixup!\" and \"squash!\" is likely\nto change the relative order of these refs, so it is not recommended.\nThe workflow here is intended to allow these kinds of commits at the tip\nof the rebased branch while the other sub branches come along for the\nride without intervention.\n\nBe careful to not attempt updating any branch that is checked out. The\nmost common example is the branch being rebased is checked out and\ndecorates the tip commit. If the user is rebasing commits reachable from\na different branch that is checked out in a different worktree, then\nthey may be surprised to not see that ref update. However, it's probably\nbest to not optimize for this scenario and do the safest thing that will\nresult in a successful rebase. A comment is left in the TODO list that\nsignals that these refs are currently checked out.\n\nSigned-off-by: Derrick Stolee \n---\n Documentation/git-rebase.txt | 7 +++\n builtin/rebase.c | 5 ++\n sequencer.c | 99 +++++++++++++++++++++++++++++++++++\n sequencer.h | 1 +\n t/t3404-rebase-interactive.sh | 29 ++++++++++\n 5 files changed, 141 insertions(+)\n\ndiff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt\nindex 262fb01aec0..866554fc978 100644\n--- a/Documentation/git-rebase.txt\n+++ b/Documentation/git-rebase.txt\n@@ -609,6 +609,13 @@ provided. Otherwise an explicit `--no-reschedule-failed-exec` at the\n start would be overridden by the presence of\n `rebase.rescheduleFailedExec=true` configuration.\n \n+--update-refs::\n+--no-update-refs::\n+\tAutomatically force-update any branches that point to commits that\n+\tare being rebased. Any branches that are checked out in a worktree\n+\tor point to a `squash! ...` or `fixup! ...` commit are not updated\n+\tin this way.\n+\n INCOMPATIBLE OPTIONS\n --------------------\n \ndiff --git a/builtin/rebase.c b/builtin/rebase.c\nindex 7ab50cda2ad..56d82a52106 100644\n--- a/builtin/rebase.c\n+++ b/builtin/rebase.c\n@@ -102,6 +102,7 @@ struct rebase_options {\n \tint reschedule_failed_exec;\n \tint reapply_cherry_picks;\n \tint fork_point;\n+\tint update_refs;\n };\n \n #define REBASE_OPTIONS_INIT {\t\t\t \t\\\n@@ -298,6 +299,7 @@ static int do_interactive_rebase(struct rebase_options *opts, unsigned flags)\n \t\tret = complete_action(the_repository, &replay, flags,\n \t\t\tshortrevisions, opts->onto_name, opts->onto,\n \t\t\t&opts->orig_head, &commands, opts->autosquash,\n+\t\t\topts->update_refs,\n \t\t\t&todo_list);\n \t}\n \n@@ -1124,6 +1126,9 @@ int cmd_rebase(int argc, const char **argv, const char *prefix)\n \t\tOPT_BOOL(0, \"autosquash\", &options.autosquash,\n \t\t\t N_(\"move commits that begin with \"\n \t\t\t \"squash!/fixup! under -i\")),\n+\t\tOPT_BOOL(0, \"update-refs\", &options.update_refs,\n+\t\t\t N_(\"update local refs that point to commits \"\n+\t\t\t \"that are being rebased\")),\n \t\t{ OPTION_STRING, 'S', \"gpg-sign\", &gpg_sign, N_(\"key-id\"),\n \t\t\tN_(\"GPG-sign commits\"),\n \t\t\tPARSE_OPT_OPTARG, NULL, (intptr_t) \"\" },\ndiff --git a/sequencer.c b/sequencer.c\nindex 8c3ed3532ac..d6151af9849 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -35,6 +35,8 @@\n #include \"commit-reach.h\"\n #include \"rebase-interactive.h\"\n #include \"reset.h\"\n+#include \"branch.h\"\n+#include \"log-tree.h\"\n \n #define GIT_REFLOG_ACTION \"GIT_REFLOG_ACTION\"\n \n@@ -5603,10 +5605,104 @@ static int skip_unnecessary_picks(struct repository *r,\n \treturn 0;\n }\n \n+struct todo_add_branch_context {\n+\tstruct todo_list new_list;\n+\tstruct strbuf *buf;\n+\tstruct commit *commit;\n+};\n+\n+static int add_branch_for_decoration(const struct name_decoration *d, void *data)\n+{\n+\tstruct todo_add_branch_context *ctx = data;\n+\tsize_t base_offset = ctx->buf->len;\n+\tint i = ctx->new_list.nr;\n+\tstruct todo_item *item;\n+\tchar *path;\n+\n+\tALLOC_GROW(ctx->new_list.items,\n+\t\t ctx->new_list.nr + 1,\n+\t\t ctx->new_list.alloc);\n+\titem = &ctx->new_list.items[i];\n+\n+\t/* If the branch is checked out, then leave a comment instead. */\n+\tif (branch_checked_out(d->name, &path)) {\n+\t\titem->command = TODO_COMMENT;\n+\t\tstrbuf_addf(ctx->buf, \"# Ref %s checked out at '%s'\\n\",\n+\t\t\t d->name, path);\n+\t\tfree(path);\n+\t} else {\n+\t\titem->command = TODO_EXEC;\n+\t\tstrbuf_addf(ctx->buf, \"git update-ref %s HEAD %s\\n\",\n+\t\t\t d->name, oid_to_hex(&ctx->commit->object.oid));\n+\t}\n+\n+\titem->commit = NULL;\n+\titem->offset_in_buf = base_offset;\n+\titem->arg_offset = base_offset;\n+\titem->arg_len = ctx->buf->len - base_offset;\n+\tctx->new_list.nr++;\n+\n+\treturn 0;\n+}\n+\n+/*\n+ * For each 'pick' command, find out if the commit has a decoration in\n+ * refs/heads/. If so, then add a 'git branch -f' exec command after\n+ * that 'pick' (plus any following 'squash' or 'fixup' commands).\n+ */\n+static int todo_list_add_branch_commands(struct todo_list *todo_list)\n+{\n+\tint i;\n+\tstatic struct string_list decorate_refs_exclude = STRING_LIST_INIT_NODUP;\n+\tstatic struct string_list decorate_refs_exclude_config = STRING_LIST_INIT_NODUP;\n+\tstatic struct string_list decorate_refs_include = STRING_LIST_INIT_NODUP;\n+\tstruct decoration_filter decoration_filter = {\n+\t\t.include_ref_pattern = &decorate_refs_include,\n+\t\t.exclude_ref_pattern = &decorate_refs_exclude,\n+\t\t.exclude_ref_config_pattern = &decorate_refs_exclude_config\n+\t};\n+\tstruct todo_add_branch_context ctx = {\n+\t\t.new_list = TODO_LIST_INIT,\n+\t\t.buf = &todo_list->buf,\n+\t};\n+\n+\tstring_list_append(&decorate_refs_include, \"refs/heads/\");\n+\tload_ref_decorations(&decoration_filter, 0);\n+\n+\tfor (i = 0; i < todo_list->nr; ) {\n+\t\tstruct todo_item *item = &todo_list->items[i];\n+\n+\t\tdo {\n+\t\t\t/* insert ith item into new list */\n+\t\t\tALLOC_GROW(ctx.new_list.items,\n+\t\t\t\t ctx.new_list.nr + 1,\n+\t\t\t\t ctx.new_list.alloc);\n+\n+\t\t\tmemcpy(&ctx.new_list.items[ctx.new_list.nr++],\n+\t\t\t &todo_list->items[i],\n+\t\t\t sizeof(struct todo_item));\n+\n+\t\t\ti++;\n+\t\t} while (i < todo_list->nr &&\n+\t\t\t todo_list->items[i].command != TODO_PICK);\n+\n+\t\tctx.commit = item->commit;\n+\t\tfor_each_decoration(item->commit, add_branch_for_decoration, &ctx);\n+\t}\n+\n+\tfree(todo_list->items);\n+\ttodo_list->items = ctx.new_list.items;\n+\ttodo_list->nr = ctx.new_list.nr;\n+\ttodo_list->alloc = ctx.new_list.alloc;\n+\n+\treturn 0;\n+}\n+\n int complete_action(struct repository *r, struct replay_opts *opts, unsigned flags,\n \t\t const char *shortrevisions, const char *onto_name,\n \t\t struct commit *onto, const struct object_id *orig_head,\n \t\t struct string_list *commands, unsigned autosquash,\n+\t\t unsigned keep_decorations,\n \t\t struct todo_list *todo_list)\n {\n \tchar shortonto[GIT_MAX_HEXSZ + 1];\n@@ -5628,6 +5724,9 @@ int complete_action(struct repository *r, struct replay_opts *opts, unsigned fla\n \tif (autosquash && todo_list_rearrange_squash(todo_list))\n \t\treturn -1;\n \n+\tif (keep_decorations && todo_list_add_branch_commands(todo_list))\n+\t\treturn -1;\n+\n \tif (commands->nr)\n \t\ttodo_list_add_exec_commands(todo_list, commands);\n \ndiff --git a/sequencer.h b/sequencer.h\nindex da64473636b..1add3c7119f 100644\n--- a/sequencer.h\n+++ b/sequencer.h\n@@ -166,6 +166,7 @@ int complete_action(struct repository *r, struct replay_opts *opts, unsigned fla\n \t\t const char *shortrevisions, const char *onto_name,\n \t\t struct commit *onto, const struct object_id *orig_head,\n \t\t struct string_list *commands, unsigned autosquash,\n+\t\t unsigned keep_decorations,\n \t\t struct todo_list *todo_list);\n int todo_list_rearrange_squash(struct todo_list *todo_list);\n \ndiff --git a/t/t3404-rebase-interactive.sh b/t/t3404-rebase-interactive.sh\nindex f31afd4a547..38c7ef95e0e 100755\n--- a/t/t3404-rebase-interactive.sh\n+++ b/t/t3404-rebase-interactive.sh\n@@ -1743,6 +1743,35 @@ test_expect_success 'ORIG_HEAD is updated correctly' '\n \ttest_cmp_rev ORIG_HEAD test-orig-head@{1}\n '\n \n+test_expect_success '--update-refs adds git branch commands' '\n+\tgit checkout -b update-refs no-conflict-branch &&\n+\ttest_commit extra fileX &&\n+\tgit commit --amend --fixup=L &&\n+\t(\n+\t\tset_cat_todo_editor &&\n+\t\tgit branch -f base A &&\n+\t\tgit branch -f first J &&\n+\t\tgit branch -f second J &&\n+\t\tgit branch -f third L &&\n+\n+\t\ttest_must_fail git rebase -i --autosquash --update-refs primary >todo &&\n+\n+\t\tcat >expect <<-EOF &&\n+\t\tpick $(git log -1 --format=%h J) J\n+\t\texec git update-ref refs/heads/second HEAD $(git rev-parse J)\n+\t\texec git update-ref refs/heads/first HEAD $(git rev-parse J)\n+\t\tpick $(git log -1 --format=%h K) K\n+\t\tpick $(git log -1 --format=%h L) L\n+\t\tfixup $(git log -1 --format=%h update-refs) fixup! L\n+\t\texec git update-ref refs/heads/third HEAD $(git rev-parse L)\n+\t\tpick $(git log -1 --format=%h M) M\n+\t\texec git update-ref refs/heads/no-conflict-branch HEAD $(git rev-parse M)\n+\t\tEOF\n+\n+\t\ttest_cmp expect todo\n+\t)\n+'\n+\n # This must be the last test in this file\n test_expect_success '$EDITOR and friends are unchanged' '\n \ttest_editor_unchanged\n-- \ngitgitgadget\n\n"},{"id":"456580","messageId":"b99c5bf34ef45f8ac9e57ef28af01dfff7a72198.1654263472.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.git.1654263472.gitgitgadget@gmail.com","subject":"[PATCH 4/4] rebase: add rebase.updateRefs config option","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-03T13:37:52Z","receivedAt":"2022-06-03T13:38:12Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe previous change added the --update-refs command-line option. For\nusers who always want this mode, create the rebase.updateRefs config\noption which behaves the same way as rebase.autoSquash does with the\n--autosquash option.\n\nSigned-off-by: Derrick Stolee \n---\n Documentation/config/rebase.txt | 3 +++\n Documentation/git-rebase.txt | 4 ++++\n builtin/rebase.c | 5 +++++\n t/t3404-rebase-interactive.sh | 5 +++++\n 4 files changed, 17 insertions(+)\n\ndiff --git a/Documentation/config/rebase.txt b/Documentation/config/rebase.txt\nindex 8c979cb20f2..f19bd0e0407 100644\n--- a/Documentation/config/rebase.txt\n+++ b/Documentation/config/rebase.txt\n@@ -21,6 +21,9 @@ rebase.autoStash::\n \t`--autostash` options of linkgit:git-rebase[1].\n \tDefaults to false.\n \n+rebase.updateRefs::\n+\tIf set to true enable `--update-refs` option by default.\n+\n rebase.missingCommitsCheck::\n \tIf set to \"warn\", git rebase -i will print a warning if some\n \tcommits are removed (e.g. a line was deleted), however the\ndiff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt\nindex 866554fc978..cb1b498bd99 100644\n--- a/Documentation/git-rebase.txt\n+++ b/Documentation/git-rebase.txt\n@@ -615,6 +615,10 @@ start would be overridden by the presence of\n \tare being rebased. Any branches that are checked out in a worktree\n \tor point to a `squash! ...` or `fixup! ...` commit are not updated\n \tin this way.\n++\n+If the `--update-refs` option is enabled by default using the\n+configuration variable `rebase.updateRefs`, this option can be\n+used to override and disable this setting.\n \n INCOMPATIBLE OPTIONS\n --------------------\ndiff --git a/builtin/rebase.c b/builtin/rebase.c\nindex 56d82a52106..8ebc98ea505 100644\n--- a/builtin/rebase.c\n+++ b/builtin/rebase.c\n@@ -802,6 +802,11 @@ static int rebase_config(const char *var, const char *value, void *data)\n \t\treturn 0;\n \t}\n \n+\tif (!strcmp(var, \"rebase.updaterefs\")) {\n+\t\topts->update_refs = git_config_bool(var, value);\n+\t\treturn 0;\n+\t}\n+\n \tif (!strcmp(var, \"rebase.reschedulefailedexec\")) {\n \t\topts->reschedule_failed_exec = git_config_bool(var, value);\n \t\treturn 0;\ndiff --git a/t/t3404-rebase-interactive.sh b/t/t3404-rebase-interactive.sh\nindex 38c7ef95e0e..1fa9f78d40d 100755\n--- a/t/t3404-rebase-interactive.sh\n+++ b/t/t3404-rebase-interactive.sh\n@@ -1768,6 +1768,11 @@ test_expect_success '--update-refs adds git branch commands' '\n \t\texec git update-ref refs/heads/no-conflict-branch HEAD $(git rev-parse M)\n \t\tEOF\n \n+\t\ttest_cmp expect todo &&\n+\n+\t\ttest_must_fail git -c rebase.autosquash=true \\\n+\t\t\t\t -c rebase.updaterefs=true \\\n+\t\t\t\t rebase -i primary >todo &&\n \t\ttest_cmp expect todo\n \t)\n '\n-- \ngitgitgadget\n"},{"id":"456594","messageId":"xmqqpmjpheiv.fsf@gitster.g","threadId":"57949","inReplyTo":"pull.1247.git.1654263472.gitgitgadget@gmail.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-03T16:56:24Z","receivedAt":"2022-06-03T16:56:33Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"\"Derrick Stolee via GitGitGadget\" writes:\n\n> This is a feature I've wanted for quite a while. When working on the sparse\n> index topic, I created a long RFC that actually broke into three topics for\n> full review upstream. These topics were sequential, so any feedback on an\n> earlier one required updates to the later ones. I would work on the full\n> feature and use interactive rebase to update the full list of commits.\n> However, I would need to update the branches pointing to those sub-topics.\n>\n> This series adds a new --update-refs option to 'git rebase' (along with a\n> rebase.updateRefs config option) that adds 'git update-ref' commands into\n> the TODO list. This is powered by the commit decoration machinery.\n\n;-)\n"},{"id":"456598","messageId":"xmqqa6athcif.fsf@gitster.g","threadId":"57949","inReplyTo":"4f9f34876413927d819313a70fcdefcad5b35689.1654263472.git.gitgitgadget@gmail.com","subject":"Re: [PATCH 1/4] log-tree: create for_each_decoration()","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-03T17:39:52Z","receivedAt":"2022-06-03T17:39:59Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"\"Derrick Stolee via GitGitGadget\" writes:\n\n> \tconst struct name_decoration *decoration;\n> +\tstruct format_decorations_context ctx = {\n> +\t\t.sb = sb,\n> +\t\t.use_color = use_color,\n> +\t\t.prefix = prefix,\n> +\t\t.separator = separator,\n> +\t\t.suffix = suffix,\n> +\t\t.color_commit =\tdiff_get_color(use_color, DIFF_COMMIT),\n> +\t\t.color_reset = decorate_get_color(use_color, DECORATION_NONE),\n> +\t};\n> \n> \tdecoration = get_name_decoration(&commit->object);\n> \tif (!decoration)\n> \t\treturn;\n> \n> +\tctx.current_and_HEAD = current_pointed_by_HEAD(decoration);\n> \n> +\tfor_each_decoration(commit, append_decoration, &ctx);\n\nThe function for_each_decoration() that does not take decoration but\na commit felt iffy, especially because we already have called\nget_name_decoration() to obtain one for commit, and the API forces\nus to do that again at the beginning of for_each_decoration().\n\n> +\tstrbuf_addstr(sb, ctx.color_commit);\n> +\tstrbuf_addstr(sb, ctx.suffix);\n> +\tstrbuf_addstr(sb, ctx.color_reset);\n> +}\n> +\n> +int for_each_decoration(const struct commit *c, decoration_fn fn, void *data)\n> +{\n> +\tconst struct name_decoration *decoration;\n> +\n> +\tdecoration = get_name_decoration(&c->object);\n> +\twhile (decoration) {\n> +\t\tint res;\n> +\t\tif ((res = fn(decoration, data)))\n> +\t\t\treturn res;\n> \t\tdecoration = decoration->next;\n> \t}\n> +\n> +\treturn 0;\n> }\n\nWe'll know if this small waste is worth it when we see the new\ncaller(s), I guess, but even if they start from commit, allowing\nthem the same early return trick would require this piece of code:\n\n\tdecoration = get_name_decoration(&commit->object);\n\tif (!decoration)\n\t\treturn;\n\nin them, and even if they do not need it, it would be trivial for\nthem to say\n\n\tfor_each_decoration(get_name_decoration(&commit->object),\n \t\t do_whatever_they_need_to_do, &ctx);\n\nso, we may want to revisit this after we finish reading the series\nthrough. Making the iterator take name_decoration does not look\ntoo bad.\n\n> void show_decorations(struct rev_info *opt, struct commit *commit)\n> diff --git a/log-tree.h b/log-tree.h\n> index e7e4641cf83..ea07da2625b 100644\n> --- a/log-tree.h\n> +++ b/log-tree.h\n> @@ -35,4 +35,8 @@ void fmt_output_commit(struct strbuf *, struct commit *, struct rev_info *);\n> void fmt_output_subject(struct strbuf *, const char *subject, struct rev_info *);\n> void fmt_output_email_subject(struct strbuf *, struct rev_info *);\n> \n> +typedef int decoration_fn(const struct name_decoration *d,\n> +\t\t\t void *data);\n> +int for_each_decoration(const struct commit *c, decoration_fn fn, void *data);\n> +\n> #endif\n"},{"id":"456599","messageId":"55d7146d-0fc4-3c29-b9a4-72f5ff856255@github.com","threadId":"57949","inReplyTo":"xmqqa6athcif.fsf@gitster.g","subject":"Re: [PATCH 1/4] log-tree: create for_each_decoration()","fromName":"Derrick Stolee","fromEmail":"derrickstolee@github.com","sentAt":"2022-06-03T17:58:32Z","receivedAt":"2022-06-03T18:08:49Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"On 6/3/2022 1:39 PM, Junio C Hamano wrote:\n> \"Derrick Stolee via GitGitGadget\" writes:\n> \n>> \tconst struct name_decoration *decoration;\n>> +\tstruct format_decorations_context ctx = {\n>> +\t\t.sb = sb,\n>> +\t\t.use_color = use_color,\n>> +\t\t.prefix = prefix,\n>> +\t\t.separator = separator,\n>> +\t\t.suffix = suffix,\n>> +\t\t.color_commit =\tdiff_get_color(use_color, DIFF_COMMIT),\n>> +\t\t.color_reset = decorate_get_color(use_color, DECORATION_NONE),\n>> +\t};\n>> \n>> \tdecoration = get_name_decoration(&commit->object);\n>> \tif (!decoration)\n>> \t\treturn;\n>> \n>> +\tctx.current_and_HEAD = current_pointed_by_HEAD(decoration);\n>> \n>> +\tfor_each_decoration(commit, append_decoration, &ctx);\n> \n> The function for_each_decoration() that does not take decoration but\n> a commit felt iffy, especially because we already have called\n> get_name_decoration() to obtain one for commit, and the API forces\n> us to do that again at the beginning of for_each_decoration().\n\nThis is more an issue with this particular caller since it needs\nthat current_pointed_by_HEAD() information. I had considered it\nas something to include in the prototype of decoration_fn so it\ncould be incorporated into for_each_decoration(), but it ended up\nbeing overly clunky for this one consumer. I'm open to other ideas,\nthough.\n\n>> +\tstrbuf_addstr(sb, ctx.color_commit);\n>> +\tstrbuf_addstr(sb, ctx.suffix);\n>> +\tstrbuf_addstr(sb, ctx.color_reset);\n>> +}\n>> +\n>> +int for_each_decoration(const struct commit *c, decoration_fn fn, void *data)\n\nThe goal of this method was to make it super simple to iterate\nwithout doing any prep work: just load the commit and go.\n\n>> +{\n>> +\tconst struct name_decoration *decoration;\n>> +\n>> +\tdecoration = get_name_decoration(&c->object);\n>> +\twhile (decoration) {\n>> +\t\tint res;\n>> +\t\tif ((res = fn(decoration, data)))\n>> +\t\t\treturn res;\n>> \t\tdecoration = decoration->next;\n>> \t}\n>> +\n>> +\treturn 0;\n>> }\n> \n> We'll know if this small waste is worth it when we see the new\n> caller(s), I guess, but even if they start from commit, allowing\n> them the same early return trick would require this piece of code:\n> \n> \tdecoration = get_name_decoration(&commit->object);\n> \tif (!decoration)\n> \t\treturn;\n\nThe iterator returns quickly because the while loop notices a\nNULL decoration. This use in format_decorations_extended() needs\nthat check because of the current_pointed_by_HEAD(decoration)\ncall based on the result being non-NULL.\n\n> in them, and even if they do not need it, it would be trivial for\n> them to say\n> \n> \tfor_each_decoration(get_name_decoration(&commit->object),\n> \t\t do_whatever_they_need_to_do, &ctx);\n> \n> so, we may want to revisit this after we finish reading the series\n> through. Making the iterator take name_decoration does not look\n> too bad.\n\nYou are right that this would work, but it's a bit messier. We\ncould go this route to avoid the duplicate get_name_decoration()\ncalls in format_decorations_extended().\n\nThanks,\n-Stolee\n"},{"id":"456602","messageId":"YppSl4rOvVciLuXE@nand.local","threadId":"57949","inReplyTo":"pull.1247.git.1654263472.gitgitgadget@gmail.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Taylor Blau","fromEmail":"me@ttaylorr.com","sentAt":"2022-06-03T18:27:35Z","receivedAt":"2022-06-03T18:36:18Z","isPatch":true,"sender":{"key":"me@ttaylorr.com","avatar":"https://avatars.githubusercontent.com/u/301000140?v=4"},"body":"On Fri, Jun 03, 2022 at 01:37:48PM +0000, Derrick Stolee via GitGitGadget wrote:\n> This is a feature I've wanted for quite a while. When working on the sparse\n> index topic, I created a long RFC that actually broke into three topics for\n> full review upstream. These topics were sequential, so any feedback on an\n> earlier one required updates to the later ones. I would work on the full\n> feature and use interactive rebase to update the full list of commits.\n> However, I would need to update the branches pointing to those sub-topics.\n\nThis is really exciting. I'm glad that you're working on it, because I\nhave also wanted this feature a handful of times over the years.\n\nThis definitely would have come in handy when writing MIDX bitmaps,\nwhere I was often editing a local branch pointing at the final topic,\nand then trying to reconstruct all of the intermediate branches after\neach rebase. Not ever having to do that again sounds like a delight ;-).\n\n> pick 2d966282ff3 docs: document bundle URI standard\n> pick 31396e9171a remote-curl: add 'get' capability\n> pick 54c6ab70f67 bundle-uri: create basic file-copy logic\n> pick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\n> pick 6adaf842684 fetch: add --bundle-uri option\n> pick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\n> exec git update-ref refs/heads/bundle-redo/fetch HEAD 6c5840ed77e1bc41c1fe6fb7c894ceede1b8d730\n\nBut I wonder if we can or should delay these update-refs as long as\npossible. In particular: what happens if I get past this \"exec\" line (so\nthat I've already updated my bundle-redo/fetch branch to point at the\nnew thing), but decide at some later point to abort the rebase?\n\nI think users will expect us to restore bundle-redo/fetch to where it\nwas before if we end up in that case. Recovering from it manually sounds\nlike kind of a headache.\n\nWhat if instead we created labels here, and then delayed all ref updates\nto the end by replacing this with:\n\n label bundle-redo/fetch\n\nand then at the end of the todo list we'd add:\n\n exec git update-ref refs/heads/bundle-redo/fetch refs/rewritten/bundle-redo/fetch\n\nIf we do all of those ref updates in a single transaction at the end, it\nshould be much easier to roll back from if desired, and we'd avoid the\naborted-rebase problem entirely.\n\nThanks,\nTaylor\n"},{"id":"456603","messageId":"xmqqzgitfvjs.fsf@gitster.g","threadId":"57949","inReplyTo":"5f54766e1032ebf3a331516a6dd696b997bdfdd8.1654263472.git.gitgitgadget@gmail.com","subject":"Re: [PATCH 2/4] branch: add branch_checked_out() helper","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-03T18:31:35Z","receivedAt":"2022-06-03T18:37:07Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"\"Derrick Stolee via GitGitGadget\" writes:\n\n> +int branch_checked_out(const char *refname, char **path)\n> +{\n> +\tstruct worktree **worktrees = get_worktrees();\n> +\tconst struct worktree *wt = find_shared_symref(worktrees, \"HEAD\", refname);\n> +\tint result = wt && !wt->is_bare;\n> +\n> +\tif (result && path)\n> +\t\t*path = xstrdup(wt->path);\n> +\n> +\tfree_worktrees(worktrees);\n> +\treturn result;\n> +}\n\nDon't you plan to call this repeatedly from the for_each_deco\niteration? I am wondering if it should take the result of\nget_worktrees() and reuse the result of get_worktrees().\n\nThere also was another topic that was triggered by\nfind_shared_symref() being relatively heavy-weight, which suggests a\nmore involved refactoring.\n\nI wonder if we rather want to rewrite find_shared_symref() *not* to\ntake the target parameter at all, and instead introduce a new\nfunction that iterates over worktrees and report the branch that is\nchecked out (or being operated on via rebase or bisect). Then we\ncan\n\n - create a strset out of its result, i.e. set of branches that\n should not be touched;\n\n - iterate over refs that point into the history being rebased\n (using for_each_decoration()), and consult that strset to see if\n any of them is being rewritten. \n\nWith the API of find_shared_symref(), we'd need to iterate over all\nworktrees for each decoration. With such a restructuring, we can\niterate over all worktrees just once, and match the result with\ndecoration, so the problem becomes O(N)+O(M) and not O(N*M) for\nnumber of worktrees N and number of decorations M.\n"},{"id":"456619","messageId":"xmqqv8thfv56.fsf@gitster.g","threadId":"57949","inReplyTo":"55d7146d-0fc4-3c29-b9a4-72f5ff856255@github.com","subject":"Re: [PATCH 1/4] log-tree: create for_each_decoration()","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-03T18:40:21Z","receivedAt":"2022-06-03T18:40:30Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Derrick Stolee writes:\n\n>>> +\tfor_each_decoration(commit, append_decoration, &ctx);\n>> \n>> The function for_each_decoration() that does not take decoration but\n>> a commit felt iffy,...\n>> ...\n>> \tfor_each_decoration(get_name_decoration(&commit->object),\n>> \t\t do_whatever_they_need_to_do, &ctx);\n>> \n>> so, we may want to revisit this after we finish reading the series\n>> through. Making the iterator take name_decoration does not look\n>> too bad.\n>\n> You are right that this would work, but it's a bit messier. We\n> could go this route to avoid the duplicate get_name_decoration()\n> calls in format_decorations_extended().\n\nHaving a convenience function that takes a commit and let you walk\nover its decoration is OK, but calling it for_each_decoration()\nbothers me somewhat (which is where my comment started from after\nall).\n\nHopefully it will stay unambiguous, and it will stay to be\nunnecessary to name it for_each_decoration_for_commit(), as I do not\nthink we'd add \"decoration\" to things that are not commits (or if we\nadded one, we probably will call it differently from \"decoration\").\n\nSo, OK.\n"},{"id":"456623","messageId":"xmqqr145fuku.fsf@gitster.g","threadId":"57949","inReplyTo":"YppSl4rOvVciLuXE@nand.local","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-03T18:52:33Z","receivedAt":"2022-06-03T18:52:38Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Taylor Blau writes:\n\n> But I wonder if we can or should delay these update-refs as long as\n> possible. In particular: what happens if I get past this \"exec\" line (so\n> that I've already updated my bundle-redo/fetch branch to point at the\n> new thing), but decide at some later point to abort the rebase?\n>\n> I think users will expect us to restore bundle-redo/fetch to where it\n> was before if we end up in that case. Recovering from it manually sounds\n> like kind of a headache.\n\nThat is a very good safety and usability concern. I am glad\nsomebody thought of it.\n\n> What if instead we created labels here, and then delayed all ref updates\n> to the end by replacing this with:\n>\n> label bundle-redo/fetch\n>\n> and then at the end of the todo list we'd add:\n>\n> exec git update-ref refs/heads/bundle-redo/fetch refs/rewritten/bundle-redo/fetch\n>\n> If we do all of those ref updates in a single transaction at the end, it\n> should be much easier to roll back from if desired, and we'd avoid the\n> aborted-rebase problem entirely.\n\n;-)\n"},{"id":"456630","messageId":"9d2752dd-6029-e71a-4149-0641fea879fe@jeffhostetler.com","threadId":"57949","inReplyTo":"YppSl4rOvVciLuXE@nand.local","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Jeff Hostetler","fromEmail":"git@jeffhostetler.com","sentAt":"2022-06-03T19:59:00Z","receivedAt":"2022-06-03T19:59:05Z","isPatch":true,"sender":{"key":"git@jeffhostetler.com","avatar":null},"body":"\n\nOn 6/3/22 2:27 PM, Taylor Blau wrote:\n> On Fri, Jun 03, 2022 at 01:37:48PM +0000, Derrick Stolee via GitGitGadget wrote:\n>> This is a feature I've wanted for quite a while. When working on the sparse\n>> index topic, I created a long RFC that actually broke into three topics for\n>> full review upstream. These topics were sequential, so any feedback on an\n>> earlier one required updates to the later ones. I would work on the full\n>> feature and use interactive rebase to update the full list of commits.\n>> However, I would need to update the branches pointing to those sub-topics.\n> \n> This is really exciting. I'm glad that you're working on it, because I\n> have also wanted this feature a handful of times over the years.\n> \n> This definitely would have come in handy when writing MIDX bitmaps,\n> where I was often editing a local branch pointing at the final topic,\n> and then trying to reconstruct all of the intermediate branches after\n> each rebase. Not ever having to do that again sounds like a delight ;-).\n> \n>> pick 2d966282ff3 docs: document bundle URI standard\n>> pick 31396e9171a remote-curl: add 'get' capability\n>> pick 54c6ab70f67 bundle-uri: create basic file-copy logic\n>> pick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\n>> pick 6adaf842684 fetch: add --bundle-uri option\n>> pick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\n>> exec git update-ref refs/heads/bundle-redo/fetch HEAD 6c5840ed77e1bc41c1fe6fb7c894ceede1b8d730\n> \n> But I wonder if we can or should delay these update-refs as long as\n> possible. In particular: what happens if I get past this \"exec\" line (so\n> that I've already updated my bundle-redo/fetch branch to point at the\n> new thing), but decide at some later point to abort the rebase?\n> \n> I think users will expect us to restore bundle-redo/fetch to where it\n> was before if we end up in that case. Recovering from it manually sounds\n> like kind of a headache.\n> \n> What if instead we created labels here, and then delayed all ref updates\n> to the end by replacing this with:\n> \n> label bundle-redo/fetch\n> \n> and then at the end of the todo list we'd add:\n> \n> exec git update-ref refs/heads/bundle-redo/fetch refs/rewritten/bundle-redo/fetch\n> \n> If we do all of those ref updates in a single transaction at the end, it\n> should be much easier to roll back from if desired, and we'd avoid the\n> aborted-rebase problem entirely.\n> \n> Thanks,\n> Taylor\n> \n\nI agree. I could have really used this while juggling all of the\nparts of FSMonitor recently. And yes, it should write the updates\nat the bottom in case of an abort.\n\nShould this take a branch pattern/regex to limit the set of branches\nthat are updated (or offered to be updated)? For example, if I have\nan intermediate commit in the series that has 2 branch names pointing\nat it, do we want to offer to update both of them or only the one\nthat matches some pattern related to the tip? Or is it sufficient to\njust enumerate them at the bottom of the todo list and let the user\ndelete the lines they don't want?\n\nShould we actually do the update-ref's or should we write a script\nthat lets the user do it later? The latter would let us also write\nout the commands to force update the remote refs if that would be\nhelpful.\n\nThanks,\nJeff\n"},{"id":"456631","messageId":"YpppCx24rn9UTXe2@nand.local","threadId":"57949","inReplyTo":"9d2752dd-6029-e71a-4149-0641fea879fe@jeffhostetler.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Taylor Blau","fromEmail":"me@ttaylorr.com","sentAt":"2022-06-03T20:03:23Z","receivedAt":"2022-06-03T20:03:31Z","isPatch":true,"sender":{"key":"me@ttaylorr.com","avatar":"https://avatars.githubusercontent.com/u/301000140?v=4"},"body":"On Fri, Jun 03, 2022 at 03:59:00PM -0400, Jeff Hostetler wrote:\n> Should this take a branch pattern/regex to limit the set of branches\n> that are updated (or offered to be updated)? For example, if I have\n> an intermediate commit in the series that has 2 branch names pointing\n> at it, do we want to offer to update both of them or only the one\n> that matches some pattern related to the tip? Or is it sufficient to\n> just enumerate them at the bottom of the todo list and let the user\n> delete the lines they don't want?\n\nThat could be a useful feature to layer on top. I don't think it's\nrequired here, though, since (as you note) users can remove lines from\nthe update-ref invocation(s) at the bottom that they don't want to\nperform.\n\n> Should we actually do the update-ref's or should we write a script\n> that lets the user do it later? The latter would let us also write\n> out the commands to force update the remote refs if that would be\n> helpful.\n\nThat seems like it would be outside the bounds of what `rebase` should\nprovide, but others may feel differently.\n\nThanks,\nTaylor\n"},{"id":"456636","messageId":"xmqqmtetcug2.fsf@gitster.g","threadId":"57949","inReplyTo":"9d2752dd-6029-e71a-4149-0641fea879fe@jeffhostetler.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-03T21:23:41Z","receivedAt":"2022-06-03T21:23:50Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Jeff Hostetler writes:\n\n> I agree. I could have really used this while juggling all of the\n> parts of FSMonitor recently. And yes, it should write the updates\n> at the bottom in case of an abort.\n>\n> Should this take a branch pattern/regex to limit the set of branches\n> that are updated (or offered to be updated)? For example, if I have\n> an intermediate commit in the series that has 2 branch names pointing\n> at it, do we want to offer to update both of them or only the one\n> that matches some pattern related to the tip? Or is it sufficient to\n> just enumerate them at the bottom of the todo list and let the user\n> delete the lines they don't want?\n\nThe latter sounds sufficient (starting from something simpler should\nwork well in this case).\n\n> Should we actually do the update-ref's or should we write a script\n> that lets the user do it later? The latter would let us also write\n> out the commands to force update the remote refs if that would be\n> helpful.\n\nAren't we writing \"a script\" already by implementing it as an\nadditional \"exec git update-ref\" in the todo list already? I\nfound that, combined with the idea to use the \"label\" Taylor\nmentioned, was the most brilliant part of this proposal.\n\n"},{"id":"456682","messageId":"a63cba6e-d25a-ee6d-57f0-9562b6235d3b@gmail.com","threadId":"57949","inReplyTo":"YppSl4rOvVciLuXE@nand.local","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Phillip Wood","fromEmail":"phillip.wood123@gmail.com","sentAt":"2022-06-04T15:28:39Z","receivedAt":"2022-06-04T15:28:49Z","isPatch":true,"sender":{"key":"phillip.wood@dunelm.org.uk","avatar":null},"body":"On 03/06/2022 19:27, Taylor Blau wrote:\n> On Fri, Jun 03, 2022 at 01:37:48PM +0000, Derrick Stolee via GitGitGadget wrote:\n>> This is a feature I've wanted for quite a while. When working on the sparse\n>> index topic, I created a long RFC that actually broke into three topics for\n>> full review upstream. These topics were sequential, so any feedback on an\n>> earlier one required updates to the later ones. I would work on the full\n>> feature and use interactive rebase to update the full list of commits.\n>> However, I would need to update the branches pointing to those sub-topics.\n> \n> This is really exciting. I'm glad that you're working on it, because I\n> have also wanted this feature a handful of times over the years.\n\nYes, thank you Stolee. I agree this will be useful, but I wonder if \nusers will be confused that --update-refs only updates the branch heads \nthat happen to be in the todo list, rather than updating all the \nbranches that contain a rewritten commit. I think the latter is \nsomething we should try to add in the future and so we should take care \nto design this topic so that is possible.\n\n> This definitely would have come in handy when writing MIDX bitmaps,\n> where I was often editing a local branch pointing at the final topic,\n> and then trying to reconstruct all of the intermediate branches after\n> each rebase. Not ever having to do that again sounds like a delight ;-).\n> \n>> pick 2d966282ff3 docs: document bundle URI standard\n>> pick 31396e9171a remote-curl: add 'get' capability\n>> pick 54c6ab70f67 bundle-uri: create basic file-copy logic\n>> pick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\n>> pick 6adaf842684 fetch: add --bundle-uri option\n>> pick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\n>> exec git update-ref refs/heads/bundle-redo/fetch HEAD 6c5840ed77e1bc41c1fe6fb7c894ceede1b8d730\n> \n> But I wonder if we can or should delay these update-refs as long as\n> possible. In particular: what happens if I get past this \"exec\" line (so\n> that I've already updated my bundle-redo/fetch branch to point at the\n> new thing), but decide at some later point to abort the rebase?\n\nAbsolutely! There is also the question of what to do if a user skips a \ncommit that is a branch head. It is not obvious if they just want to \ndrop that commit from the branch or if they want to skip updating the \nref as well (I guess they can edit the todo list for the latter case).\n\n> I think users will expect us to restore bundle-redo/fetch to where it\n> was before if we end up in that case. Recovering from it manually sounds\n> like kind of a headache.\n> \n> What if instead we created labels here, and then delayed all ref updates\n> to the end by replacing this with:\n> \n> label bundle-redo/fetch\n\nInstead of using 'label' and 'exec' I'd prefer a new todo list command \n('update-ref' or 'update-branch'?) used in place of 'label' that takes a \nbranch name and updates the branch ref at the end of the rebase. That \nwould make it easy to do all the updates in a single transaction as you \nsuggested. Adding exec lines to do this makes the todo list messy and we \nhave been trying to stop rebase forking all the time.\n\nBest Wishes\n\nPhillip\n\n> and then at the end of the todo list we'd add:\n> \n> exec git update-ref refs/heads/bundle-redo/fetch refs/rewritten/bundle-redo/fetch\n> \n> If we do all of those ref updates in a single transaction at the end, it\n> should be much easier to roll back from if desired, and we'd avoid the\n> aborted-rebase problem entirely.\n\n\n> Thanks,\n> Taylor\n\n"},{"id":"456716","messageId":"3bb8baa2-e3e0-5cf1-aace-b0ba3ec6eb77@github.com","threadId":"57949","inReplyTo":"a63cba6e-d25a-ee6d-57f0-9562b6235d3b@gmail.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Derrick Stolee","fromEmail":"derrickstolee@github.com","sentAt":"2022-06-06T15:12:46Z","receivedAt":"2022-06-06T15:12:53Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"On 6/4/2022 11:28 AM, Phillip Wood wrote:\n> On 03/06/2022 19:27, Taylor Blau wrote:\n>> On Fri, Jun 03, 2022 at 01:37:48PM +0000, Derrick Stolee via GitGitGadget wrote:\n>>> This is a feature I've wanted for quite a while. When working on the sparse\n>>> index topic, I created a long RFC that actually broke into three topics for\n>>> full review upstream. These topics were sequential, so any feedback on an\n>>> earlier one required updates to the later ones. I would work on the full\n>>> feature and use interactive rebase to update the full list of commits.\n>>> However, I would need to update the branches pointing to those sub-topics.\n>>\n>> This is really exciting. I'm glad that you're working on it, because I\n>> have also wanted this feature a handful of times over the years.\n> \n> Yes, thank you Stolee. I agree this will be useful, but I wonder if users\n> will be confused that --update-refs only updates the branch heads that happen\n> to be in the todo list, rather than updating all the branches that contain a\n> rewritten commit. I think the latter is something we should try to add in the\n> future and so we should take care to design this topic so that is possible.\n\nAt the moment, the design adds a comment to the TODO list, showing which\nbranches are not possible to move because they are checked out at another\nworktree (or is currently checked out and will be updated by the rebase\nitself). That seems like a good place to insert alternative logic in the\nfuture if we see a need for better behavior here.\n\nUnless: am I misunderstanding something about your concern here? Are you\nworried about refs outside of refs/heads/*? Are you concerned about it\nbeing _all_ refs/heads/* that we find?\n\nOne potential way to extend this (in the future) is to make --update-refs\ntake an optional string argument containing a refspec. This would replace\nthe default refspec of refs/heads/*.\n \n>> This definitely would have come in handy when writing MIDX bitmaps,\n>> where I was often editing a local branch pointing at the final topic,\n>> and then trying to reconstruct all of the intermediate branches after\n>> each rebase. Not ever having to do that again sounds like a delight ;-).\n>>\n>>> pick 2d966282ff3 docs: document bundle URI standard\n>>> pick 31396e9171a remote-curl: add 'get' capability\n>>> pick 54c6ab70f67 bundle-uri: create basic file-copy logic\n>>> pick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\n>>> pick 6adaf842684 fetch: add --bundle-uri option\n>>> pick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\n>>> exec git update-ref refs/heads/bundle-redo/fetch HEAD 6c5840ed77e1bc41c1fe6fb7c894ceede1b8d730\n>>\n>> But I wonder if we can or should delay these update-refs as long as\n>> possible. In particular: what happens if I get past this \"exec\" line (so\n>> that I've already updated my bundle-redo/fetch branch to point at the\n>> new thing), but decide at some later point to abort the rebase?\n> \n> Absolutely! There is also the question of what to do if a user skips a\n> commit that is a branch head. It is not obvious if they just want to drop> that commit from the branch or if they want to skip updating the ref as> well (I guess they can edit the todo list for the latter case).\n\n>> I think users will expect us to restore bundle-redo/fetch to where it\n>> was before if we end up in that case. Recovering from it manually sounds\n>> like kind of a headache.\n>>\n>> What if instead we created labels here, and then delayed all ref updates\n>> to the end by replacing this with:\n>>\n>>      label bundle-redo/fetch\n> \n> Instead of using 'label' and 'exec' I'd prefer a new todo list command\n> ('update-ref' or 'update-branch'?) used in place of 'label' that takes a\n> branch name and updates the branch ref at the end of the rebase. That\n> would make it easy to do all the updates in a single transaction as you\n> suggested. Adding exec lines to do this makes the todo list messy and we\n> have been trying to stop rebase forking all the time.\n\nThanks. Yes, a new 'update-refs' step at the end would be good to make\nit clear we want to rewrite the refs in one go without a possible\ninterruption from the user.\n\nThanks,\n-Stolee\n"},{"id":"456723","messageId":"xmqqr141agwb.fsf@gitster.g","threadId":"57949","inReplyTo":"a63cba6e-d25a-ee6d-57f0-9562b6235d3b@gmail.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-06T16:36:04Z","receivedAt":"2022-06-06T16:36:15Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Phillip Wood writes:\n\n> Instead of using 'label' and 'exec' I'd prefer a new todo list command\n> ('update-ref' or 'update-branch'?) used in place of 'label' that takes\n> a branch name and updates the branch ref at the end of the\n> rebase. That would make it easy to do all the updates in a single\n> transaction as you suggested.\n\nSounds like a good approach.\n"},{"id":"456769","messageId":"CABPp-BEOV53oBoBp4YjiRfksZMmAADanZUUemhxwn7Wor=m-nA@mail.gmail.com","threadId":"57949","inReplyTo":"pull.1247.git.1654263472.gitgitgadget@gmail.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Elijah Newren","fromEmail":"newren@gmail.com","sentAt":"2022-06-07T06:25:25Z","receivedAt":"2022-06-07T06:25:44Z","isPatch":true,"sender":{"key":"newren@gmail.com","avatar":"https://avatars.githubusercontent.com/u/5455730?v=4"},"body":"On Fri, Jun 3, 2022 at 11:01 AM Derrick Stolee via GitGitGadget\n wrote:\n>\n> This is a feature I've wanted for quite a while. When working on the sparse\n> index topic, I created a long RFC that actually broke into three topics for\n> full review upstream. These topics were sequential, so any feedback on an\n> earlier one required updates to the later ones. I would work on the full\n> feature and use interactive rebase to update the full list of commits.\n> However, I would need to update the branches pointing to those sub-topics.\n>\n> This series adds a new --update-refs option to 'git rebase' (along with a\n> rebase.updateRefs config option) that adds 'git update-ref' commands into\n> the TODO list. This is powered by the commit decoration machinery.\n>\n> As an example, here is my in-progress bundle URI RFC split into subtopics as\n> they appear during the TODO list of a git rebase -i --update-refs:\n>\n> pick 2d966282ff3 docs: document bundle URI standard\n> pick 31396e9171a remote-curl: add 'get' capability\n> pick 54c6ab70f67 bundle-uri: create basic file-copy logic\n> pick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\n> pick 6adaf842684 fetch: add --bundle-uri option\n> pick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\n> exec git update-ref refs/heads/bundle-redo/fetch HEAD 6c5840ed77e1bc41c1fe6fb7c894ceede1b8d730\n>\n> pick 1e3f6546632 clone: add --bundle-uri option\n> pick 9e4a6fe9b68 clone: --bundle-uri cannot be combined with --depth\n> exec git update-ref refs/heads/bundle-redo/clone HEAD 9e4a6fe9b68a8455b427c9ac8cdbff30c96653b4\n>\n> pick 5451cb6599c bundle-uri: create bundle_list struct and helpers\n> pick 3029c3aca15 bundle-uri: create base key-value pair parsing\n> pick a8b2de79ce8 bundle-uri: create \"key=value\" line parsing\n> pick 92625a47673 bundle-uri: unit test \"key=value\" parsing\n> pick a8616af4dc2 bundle-uri: limit recursion depth for bundle lists\n> pick 9d6809a8d53 bundle-uri: parse bundle list in config format\n> pick 287a732b54c bundle-uri: fetch a list of bundles\n> exec git update-ref refs/heads/bundle-redo/list HEAD 287a732b54c4d95e7f410b3b36ef90d8a19cd346\n>\n> pick b09f8226185 protocol v2: add server-side \"bundle-uri\" skeleton\n> pick 520204dcd1c bundle-uri client: add minimal NOOP client\n> pick 62e8b457b48 bundle-uri client: add \"git ls-remote-bundle-uri\"\n> pick 00eae925043 bundle-uri: serve URI advertisement from bundle.* config\n> pick 4277440a250 bundle-uri client: add boolean transfer.bundleURI setting\n> pick caf4599a81d bundle-uri: allow relative URLs in bundle lists\n> pick df255000b7e bundle-uri: download bundles from an advertised list\n> pick d71beabf199 clone: unbundle the advertised bundles\n> pick c9578391976 t5601: basic bundle URI tests\n> # Ref refs/heads/bundle-redo/rfc-3 checked out at '/home/stolee/_git/git-bundles'\n>\n> exec git update-ref refs/heads/bundle-redo/advertise HEAD c9578391976ab9899c4e4f9b5fa2827650097305\n>\n>\n> The first two patches are helpers that are needed, but the full logic of the\n> --update-refs option is introduced in patch 3. The config option is\n> available in patch 4.\n\nVery interesting; nice to see that others are thinking on a similar\nwavelength. I've got some related work in git-replay, and it heavily\nuses update-refs style handling.\n\nNot sure if there are any differences worth affecting your design or\nmine, but perhaps it's worth mentioning my current work and plans in\ngit-replay related to the use of update-refs:\n\n\n\nSince git-replay doesn't touch the working tree or index, I had the\nnatural question of what the output should be and whether ref-updates\nshould automatically happen. I decided that like git-merge-tree, ref\nupdates should not be automatic (and thus it behaves somewhat like a\n\"dry run\" version of rebase/cherry-pick). Since I also want to\nintrinsically handle replaying multiple branches simultaneously, the\nupdate-ref style output seemed like a natural fit to me (meaning users\nwould pipe the output from git-reply to git-update-ref). Thus, one\ncan do:\n\n $ git replay --onto main startpoint..mytopic\n\nand see git-update-ref style output:\n\n update refs/heads/mytopic ${NEW_mytopic_HASH} ${OLD_mytopic_HASH}\n\nwhere ${NEW_mytopic_HASH} is the commit at the tip of a chain of\nrebased mytopic commits. (Sidenote: There's the question of whether\nto update 'main' (i.e. cherry-picking) or 'mytopic' (i.e. rebasing);\nthus, users have to choose --advance vs. --onto to specify which of\nthese should happen, except in cases where it can be implicitly\ndetermined.)\n\nOr, if you have several branches to update:\n\n $ git replay --onto main ^startpoint mytopic1 mytopic2 mytopic3\n\nand see\n\n update refs/heads/mytopic1 ${NEW_mytopic1_HASH} ${OLD_mytopic1_HASH}\n update refs/heads/mytopic2 ${NEW_mytopic2_HASH} ${OLD_mytopic2_HASH}\n update refs/heads/mytopic3 ${NEW_mytopic3_HASH} ${OLD_mytopic3_HASH}\n\n(and yes, common commits from before will be replayed and then shared\nby the new branch versions.)\n\nIf the topics are totally contained within each other, you can do this\nmore simply as\n\n $ git replay --contained --onto main main..my_tip_topic\n\nand still see all the update commands:\n\n update refs/heads/mytopic1 ${NEW_mytopic1_HASH} ${OLD_mytopic1_HASH}\n update refs/heads/mytopic2 ${NEW_mytopic2_HASH} ${OLD_mytopic2_HASH}\n update refs/heads/mytopic3 ${NEW_mytopic3_HASH} ${OLD_mytopic3_HASH}\n update refs/heads/my_tip_topic ${NEW_my_tip_topic_HASH}\n${OLD_my_tip_topic_HASH}\n\nThe option to list several refs provides a natural way of allowing\nusers to control what gets updated without updating all contained\nbranches, though rebase's implicit ranges doesn't really have a nice\nanalogue for you due to its implicit range handling.\n\nRight now, git-replay doesn't handle conflicts (it just die()s). That\nand a few other thing are still TODO.\n\nHowever, git-replay does already handle basic replaying of merges,\nincluding ones with evil changes (so long as none of the merges\ninvolved have conflicts). So, I'm hoping git-replay will also useful\nfor stuff like:\n\n $ git replay --first-parent --interactive --onto next ^next\nspecial_topic seen\n\nwhere special_topic is one of the topics previously merged into seen.\nThis last commit would allow seen to be rebuilt with some changes to\nspecial_topic. Basically, it makes whatever interactive changes are\nspecified to special_topic followed by replaying all the merges in\nnext..seen (including any necessary conflict resolutions or other\nsemantic changes recorded in the merge commits being replayed), unless\nthose . However, I do need to figure out some syntax for keeping\nspecial_topic from being transplanted onto next here...\n"},{"id":"456780","messageId":"90ccd923-3552-fe88-5d6d-869def7f1aeb@gmail.com","threadId":"57949","inReplyTo":"3bb8baa2-e3e0-5cf1-aace-b0ba3ec6eb77@github.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Phillip Wood","fromEmail":"phillip.wood123@gmail.com","sentAt":"2022-06-07T10:11:56Z","receivedAt":"2022-06-07T10:12:06Z","isPatch":true,"sender":{"key":"phillip.wood@dunelm.org.uk","avatar":null},"body":"On 06/06/2022 16:12, Derrick Stolee wrote:\n> On 6/4/2022 11:28 AM, Phillip Wood wrote:\n>> On 03/06/2022 19:27, Taylor Blau wrote:\n>>> On Fri, Jun 03, 2022 at 01:37:48PM +0000, Derrick Stolee via GitGitGadget wrote:\n>>>> This is a feature I've wanted for quite a while. When working on the sparse\n>>>> index topic, I created a long RFC that actually broke into three topics for\n>>>> full review upstream. These topics were sequential, so any feedback on an\n>>>> earlier one required updates to the later ones. I would work on the full\n>>>> feature and use interactive rebase to update the full list of commits.\n>>>> However, I would need to update the branches pointing to those sub-topics.\n>>>\n>>> This is really exciting. I'm glad that you're working on it, because I\n>>> have also wanted this feature a handful of times over the years.\n>>\n>> Yes, thank you Stolee. I agree this will be useful, but I wonder if users\n>> will be confused that --update-refs only updates the branch heads that happen\n>> to be in the todo list, rather than updating all the branches that contain a\n>> rewritten commit. I think the latter is something we should try to add in the\n>> future and so we should take care to design this topic so that is possible.\n> \n> At the moment, the design adds a comment to the TODO list, showing which\n> branches are not possible to move because they are checked out at another\n> worktree (or is currently checked out and will be updated by the rebase\n> itself). That seems like a good place to insert alternative logic in the\n> future if we see a need for better behavior here.\n\nI think the question of whether to update branches that are checked out \nin another worktree is a question of whether it is less inconvenient to \nthe user to skip the ref update and leave the user to manually update \nthe branch or to update the ref and leave the worktree in a potentially \nawkward state if the user was half way through building a commit. The \nanswer probably depends on the preferences of the user.\n\nI've been using a script that updates the refs for all the branches \nbeing rewritten for a while and have found it preferable to always \nupdate the ref rather than have to do it manually. My script also \nupdates the worktree checkout to the new HEAD if there are no \nuncommitted changes which I have found very convenient. My preference is \nprobably because I tend not to have uncommitted changes lying around in \nthe worktrees whose branches get updated.\n\n> Unless: am I misunderstanding something about your concern here? Are you\n> worried about refs outside of refs/heads/*? Are you concerned about it\n> being _all_ refs/heads/* that we find?\n\nMy concerns are primarily about being able to extend the --update-ref \noption in a backwards compatible way.\n\nI'd like to be able to add functionality to rebase all refs/heads/* that \nare descended from the commits that a simple rebase would rewrite. Say I \nwant to edit $commit then I want the rebase to rewrite all the commits \nand update all the branches in in\n\n git rev-list $(git for-each-ref --contains $commit refs/heads/) \n^$commit^@\n\nIdeally we'd avoid adding a new commandline option when adding that. I \nthink we could use an optional argument as you suggest below (though it \nwould not be a refspec).\n\n> One potential way to extend this (in the future) is to make --update-refs\n> take an optional string argument containing a refspec. This would replace\n> the default refspec of refs/heads/*.\n> [...] \n>> Instead of using 'label' and 'exec' I'd prefer a new todo list command\n>> ('update-ref' or 'update-branch'?) used in place of 'label' that takes a\n>> branch name and updates the branch ref at the end of the rebase. That\n>> would make it easy to do all the updates in a single transaction as you\n>> suggested. Adding exec lines to do this makes the todo list messy and we\n>> have been trying to stop rebase forking all the time.\n> \n> Thanks. Yes, a new 'update-refs' step at the end would be good to make\n> it clear we want to rewrite the refs in one go without a possible\n> interruption from the user.\n\nThat's great, it is a bit more work for you but I think it gives a much \nnicer UI.\n\nBest Wishes\n\nPhillip\n\n> Thanks,\n> -Stolee\n"},{"id":"456781","messageId":"c9a5369a-8c87-0ee8-bed7-16b2cd58717b@gmail.com","threadId":"57949","inReplyTo":"23fa663886405596370e5fc95f679ef299f0725c.1654263472.git.gitgitgadget@gmail.com","subject":"Re: [PATCH 3/4] rebase: add --update-refs option","fromName":"Phillip Wood","fromEmail":"phillip.wood123@gmail.com","sentAt":"2022-06-07T10:25:01Z","receivedAt":"2022-06-07T10:25:15Z","isPatch":true,"sender":{"key":"phillip.wood@dunelm.org.uk","avatar":null},"body":"Hi Stolee\n\nJust a couple of minor comments.\n\nOn 03/06/2022 14:37, Derrick Stolee via GitGitGadget wrote:\n> From: Derrick Stolee \n> [...]\n> diff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt\n> index 262fb01aec0..866554fc978 100644\n> --- a/Documentation/git-rebase.txt\n> +++ b/Documentation/git-rebase.txt\n> @@ -609,6 +609,13 @@ provided. Otherwise an explicit `--no-reschedule-failed-exec` at the\n> start would be overridden by the presence of\n> `rebase.rescheduleFailedExec=true` configuration.\n> \n> +--update-refs::\n> +--no-update-refs::\n> +\tAutomatically force-update any branches that point to commits that\n> +\tare being rebased. Any branches that are checked out in a worktree\n> +\tor point to a `squash! ...` or `fixup! ...` commit are not updated\n> +\tin this way.\n> +\n> INCOMPATIBLE OPTIONS\n> --------------------\n\nWe should add --update-refs to the list of options that are incompatible \nwith --apply.\n\n> diff --git a/sequencer.c b/sequencer.c\n> index 8c3ed3532ac..d6151af9849 100644\n> --- a/sequencer.c\n> +++ b/sequencer.c\n> @@ -35,6 +35,8 @@\n> #include \"commit-reach.h\"\n> #include \"rebase-interactive.h\"\n> #include \"reset.h\"\n> +#include \"branch.h\"\n> +#include \"log-tree.h\"\n> \n> #define GIT_REFLOG_ACTION \"GIT_REFLOG_ACTION\"\n> \n> @@ -5603,10 +5605,104 @@ static int skip_unnecessary_picks(struct repository *r,\n> \treturn 0;\n> }\n> \n> +struct todo_add_branch_context {\n> +\tstruct todo_list new_list;\n\nRather than using a struct todo_list I think it would be simpler overall \nto add\n\tstruct todo_list_item *item;\n\tsize_t nr, alloc;\n\ninstead, as I found it confusing that we were (correctly) using the \nstrbuf of the old list when adding the update-ref line to the new list.\n\n> +\tstruct strbuf *buf;\n> +\tstruct commit *commit;\n> +};\n> [...]\n> +\tfor (i = 0; i < todo_list->nr; ) {\n> +\t\tstruct todo_item *item = &todo_list->items[i];\n> +\n> +\t\tdo {\n> +\t\t\t/* insert ith item into new list */\n> +\t\t\tALLOC_GROW(ctx.new_list.items,\n> +\t\t\t\t ctx.new_list.nr + 1,\n> +\t\t\t\t ctx.new_list.alloc);\n> +\n> +\t\t\tmemcpy(&ctx.new_list.items[ctx.new_list.nr++],\n> +\t\t\t &todo_list->items[i],\n> +\t\t\t sizeof(struct todo_item));\n\nMay be\n\tctx.new_list.items[ctx.new_list.nr++] = todo_list->items[i++];\nwould be clearer\n\nBest Wishes\n\nPhillip\n"},{"id":"456817","messageId":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.git.1654263472.gitgitgadget@gmail.com","subject":"[PATCH v2 0/7] rebase: update branches in multi-part topic","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-07T20:42:42Z","receivedAt":"2022-06-08T00:27:58Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"This is a feature I've wanted for quite a while. When working on the sparse\nindex topic, I created a long RFC that actually broke into three topics for\nfull review upstream. These topics were sequential, so any feedback on an\nearlier one required updates to the later ones. I would work on the full\nfeature and use interactive rebase to update the full list of commits.\nHowever, I would need to update the branches pointing to those sub-topics.\n\nThis series adds a new --update-refs option to 'git rebase' (along with a\nrebase.updateRefs config option) that adds 'label' and 'update-refs'\ncommands into the TODO list. This is powered by the commit decoration\nmachinery.\n\nAs an example, here is my in-progress bundle URI RFC split into subtopics as\nthey appear during the TODO list of a git rebase -i --update-refs:\n\npick 2d966282ff3 docs: document bundle URI standard\npick 31396e9171a remote-curl: add 'get' capability\npick 54c6ab70f67 bundle-uri: create basic file-copy logic\npick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\npick 6adaf842684 fetch: add --bundle-uri option\npick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\nlabel for-update-refs/refs/heads/bundle-redo/fetch\n\npick 1e3f6546632 clone: add --bundle-uri option\npick 9e4a6fe9b68 clone: --bundle-uri cannot be combined with --depth\nlabel for-update-refs/refs/heads/bundle-redo/clone\n\npick 5451cb6599c bundle-uri: create bundle_list struct and helpers\npick 3029c3aca15 bundle-uri: create base key-value pair parsing\npick a8b2de79ce8 bundle-uri: create \"key=value\" line parsing\npick 92625a47673 bundle-uri: unit test \"key=value\" parsing\npick a8616af4dc2 bundle-uri: limit recursion depth for bundle lists\npick 9d6809a8d53 bundle-uri: parse bundle list in config format\npick 287a732b54c bundle-uri: fetch a list of bundles\nlabel for-update-refs/refs/heads/bundle-redo/list\n\npick b09f8226185 protocol v2: add server-side \"bundle-uri\" skeleton\npick 520204dcd1c bundle-uri client: add minimal NOOP client\npick 62e8b457b48 bundle-uri client: add \"git ls-remote-bundle-uri\"\npick 00eae925043 bundle-uri: serve URI advertisement from bundle.* config\npick 4277440a250 bundle-uri client: add boolean transfer.bundleURI setting\npick caf4599a81d bundle-uri: allow relative URLs in bundle lists\npick df255000b7e bundle-uri: download bundles from an advertised list\npick d71beabf199 clone: unbundle the advertised bundles\npick c9578391976 t5601: basic bundle URI tests\n# Ref refs/heads/bundle-redo/rfc-3 checked out at '/home/stolee/_git/git-bundles'\n\nlabel for-update-refs/refs/heads/bundle-redo/advertise\n\nupdate-refs\n\n\n * Patches 1-2 are basic refactors that we will need for our implementation.\n * Patch 3 is a cleanup of an array with indexes given by the todo_command\n enum.\n * Patches 4-6 implement the feature incrementally. For instance, we can\n test the edits to the TODO file in patch 5, but we don't actually do\n anything when seeing the 'update-refs' command until patch 6.\n * Patch 7 adds the rebase.updateRefs config option similar to\n rebase.autoSquash.\n\n\nUpdates in v2\n=============\n\nAs recommended by the excellent feedback, I have removed the 'exec' commands\nin favor of the 'label' commands and a new 'update-refs' command at the very\nend. This way, there is only one step that updates all of the refs at the\nend instead of updating refs during the rebase. If a user runs 'git rebase\n--abort' in the middle, then their refs are still where they need to be.\n\nBased on some of the discussion, it seemed like one way to do this would be\nto have an 'update-ref ' command that would take the place of these 'label'\ncommands. However, this would require two things that make it a bit awkward:\n\n 1. We would need to replicate the storage of those positions during the\n rebase. 'label' already does this pretty well. I've added the\n \"for-update-refs/\" label to help here.\n 2. If we want to close out all of the refs as the rebase is finishing, then\n that \"step\" becomes invisible to the user (and a bit more complicated to\n insert). Thus, the 'update-refs' step performs this action. If the user\n wants to do things after that step, then they can do so by editing the\n TODO list.\n\nOther updates:\n\n * The 'keep_decorations' parameter was renamed to 'update_refs'.\n * I added tests for --rebase-merges=rebase-cousins to show how these labels\n interact with other labels and merge commands.\n * I changed the order of the insertion of these update-refs labels to be\n before the fixups are rearranged. This fixes a bug where the tip commit\n is a fixup! so its decorations are never inspected (and they would be in\n the wrong place even if they were). The fixup! commands are properly\n inserted between a pick and its following label command. Tests\n demonstrate this is correct.\n * Numerous style choices are updated based on feedback.\n\nThank you for all of the detailed review and ideas in this space. I\nappreciate any more ideas that can make this feature as effective as it can\nbe.\n\nThanks, -Stolee\n\nDerrick Stolee (7):\n log-tree: create for_each_decoration()\n branch: add branch_checked_out() helper\n sequencer: define array with enum values\n sequencer: add update-refs command\n rebase: add --update-refs option\n sequencer: implement 'update-refs' command\n rebase: add rebase.updateRefs config option\n\n Documentation/config/rebase.txt | 3 +\n Documentation/git-rebase.txt | 12 ++\n branch.c | 24 ++--\n branch.h | 8 ++\n builtin/rebase.c | 10 ++\n log-tree.c | 111 +++++++++++------\n log-tree.h | 4 +\n sequencer.c | 206 +++++++++++++++++++++++++++++---\n sequencer.h | 2 +\n t/t3404-rebase-interactive.sh | 110 +++++++++++++++++\n 10 files changed, 432 insertions(+), 58 deletions(-)\n\n\nbase-commit: 2668e3608e47494f2f10ef2b6e69f08a84816bcb\nPublished-As: https://github.com/gitgitgadget/git/releases/tag/pr-1247%2Fderrickstolee%2Frebase-keep-decorations-v2\nFetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1247/derrickstolee/rebase-keep-decorations-v2\nPull-Request: https://github.com/gitgitgadget/git/pull/1247\n\nRange-diff vs v1:\n\n 1: 4f9f3487641 = 1: 4f9f3487641 log-tree: create for_each_decoration()\n 2: 5f54766e103 = 2: 5f54766e103 branch: add branch_checked_out() helper\n -: ----------- > 3: 9f261c7df2c sequencer: define array with enum values\n -: ----------- > 4: 842b2186d25 sequencer: add update-refs command\n 3: 23fa6638864 ! 5: 0a4c110127b rebase: add --update-refs option\n @@ Commit message\n reachable from those \"sub branches\". It can take a manual step to update\n those branches.\n \n - Add a new --update-refs option to 'git rebase -i' that adds 'git\n - update-ref' exec steps to the todo file whenever a commit that is being\n - rebased is decorated with that . This allows the user to rebase a\n - long list of commits in a multi-part feature and keep all of their\n - pointers to those parts.\n + Add a new --update-refs option to 'git rebase -i' that adds 'label\n + for-update-refs/*' steps to the todo file whenever a commit that is\n + being rebased is decorated with that . At the very end, the\n + 'update-refs' step is added to update all of the branches referenced by\n + the 'label' steps. This allows the user to rebase a long list of commits\n + in a multi-part feature and keep all of their pointers to those parts.\n +\n + NOTE: This change only introduce the --update-refs option and implements\n + the changes to the todo file. It does _not_ yet implement the action\n + taken by the 'update-refs' todo step, which will be implemented and\n + tested in a later change.\n \n Use the new for_each_decoration() while iterating over the existing todo\n list. Be sure to iterate after any squashing or fixups are placed.\n @@ Documentation/git-rebase.txt: provided. Otherwise an explicit `--no-reschedule-f\n INCOMPATIBLE OPTIONS\n --------------------\n \n +@@ Documentation/git-rebase.txt: are incompatible with the following options:\n + * --empty=\n + * --reapply-cherry-picks\n + * --edit-todo\n ++ * --update-refs\n + * --root when used in combination with --onto\n + \n + In addition, the following pairs of options are incompatible:\n \n ## builtin/rebase.c ##\n @@ builtin/rebase.c: struct rebase_options {\n @@ sequencer.c: static int skip_unnecessary_picks(struct repository *r,\n }\n \n +struct todo_add_branch_context {\n -+\tstruct todo_list new_list;\n ++\tstruct todo_item *items;\n ++\tsize_t items_nr;\n ++\tsize_t items_alloc;\n +\tstruct strbuf *buf;\n +\tstruct commit *commit;\n +};\n @@ sequencer.c: static int skip_unnecessary_picks(struct repository *r,\n +{\n +\tstruct todo_add_branch_context *ctx = data;\n +\tsize_t base_offset = ctx->buf->len;\n -+\tint i = ctx->new_list.nr;\n +\tstruct todo_item *item;\n +\tchar *path;\n +\n -+\tALLOC_GROW(ctx->new_list.items,\n -+\t\t ctx->new_list.nr + 1,\n -+\t\t ctx->new_list.alloc);\n -+\titem = &ctx->new_list.items[i];\n ++\tALLOC_GROW(ctx->items,\n ++\t\t ctx->items_nr + 1,\n ++\t\t ctx->items_alloc);\n ++\titem = &ctx->items[ctx->items_nr];\n ++\tmemset(item, 0, sizeof(*item));\n +\n +\t/* If the branch is checked out, then leave a comment instead. */\n +\tif (branch_checked_out(d->name, &path)) {\n @@ sequencer.c: static int skip_unnecessary_picks(struct repository *r,\n +\t\t\t d->name, path);\n +\t\tfree(path);\n +\t} else {\n -+\t\titem->command = TODO_EXEC;\n -+\t\tstrbuf_addf(ctx->buf, \"git update-ref %s HEAD %s\\n\",\n -+\t\t\t d->name, oid_to_hex(&ctx->commit->object.oid));\n ++\t\titem->command = TODO_LABEL;\n ++\t\tstrbuf_addf(ctx->buf, \"for-update-refs/%s\\n\", d->name);\n +\t}\n +\n -+\titem->commit = NULL;\n +\titem->offset_in_buf = base_offset;\n +\titem->arg_offset = base_offset;\n +\titem->arg_len = ctx->buf->len - base_offset;\n -+\tctx->new_list.nr++;\n ++\tctx->items_nr++;\n +\n +\treturn 0;\n +}\n +\n +/*\n + * For each 'pick' command, find out if the commit has a decoration in\n -+ * refs/heads/. If so, then add a 'git branch -f' exec command after\n -+ * that 'pick' (plus any following 'squash' or 'fixup' commands).\n ++ * refs/heads/. If so, then add a 'label for-update-refs/' command.\n + */\n -+static int todo_list_add_branch_commands(struct todo_list *todo_list)\n ++static int todo_list_add_update_ref_commands(struct todo_list *todo_list)\n +{\n +\tint i;\n +\tstatic struct string_list decorate_refs_exclude = STRING_LIST_INIT_NODUP;\n @@ sequencer.c: static int skip_unnecessary_picks(struct repository *r,\n +\t\t.exclude_ref_config_pattern = &decorate_refs_exclude_config\n +\t};\n +\tstruct todo_add_branch_context ctx = {\n -+\t\t.new_list = TODO_LIST_INIT,\n +\t\t.buf = &todo_list->buf,\n +\t};\n +\n ++\tctx.items_alloc = 2 * todo_list->nr + 1;\n ++\tALLOC_ARRAY(ctx.items, ctx.items_alloc);\n ++\n +\tstring_list_append(&decorate_refs_include, \"refs/heads/\");\n +\tload_ref_decorations(&decoration_filter, 0);\n +\n +\tfor (i = 0; i < todo_list->nr; ) {\n +\t\tstruct todo_item *item = &todo_list->items[i];\n +\n -+\t\tdo {\n -+\t\t\t/* insert ith item into new list */\n -+\t\t\tALLOC_GROW(ctx.new_list.items,\n -+\t\t\t\t ctx.new_list.nr + 1,\n -+\t\t\t\t ctx.new_list.alloc);\n ++\t\t/* insert ith item into new list */\n ++\t\tALLOC_GROW(ctx.items,\n ++\t\t\t ctx.items_nr + 1,\n ++\t\t\t ctx.items_alloc);\n +\n -+\t\t\tmemcpy(&ctx.new_list.items[ctx.new_list.nr++],\n -+\t\t\t &todo_list->items[i],\n -+\t\t\t sizeof(struct todo_item));\n ++\t\tctx.items[ctx.items_nr++] = todo_list->items[i++];\n +\n -+\t\t\ti++;\n -+\t\t} while (i < todo_list->nr &&\n -+\t\t\t todo_list->items[i].command != TODO_PICK);\n -+\n -+\t\tctx.commit = item->commit;\n -+\t\tfor_each_decoration(item->commit, add_branch_for_decoration, &ctx);\n ++\t\tif (item->commit) {\n ++\t\t\tctx.commit = item->commit;\n ++\t\t\tfor_each_decoration(item->commit,\n ++\t\t\t\t\t add_branch_for_decoration,\n ++\t\t\t\t\t &ctx);\n ++\t\t}\n +\t}\n +\n ++\t/* Add the \"update-refs\" step. */\n ++\tALLOC_GROW(ctx.items,\n ++\t\t ctx.items_nr + 1,\n ++\t\t ctx.items_alloc);\n ++\tmemset(&ctx.items[ctx.items_nr], 0, sizeof(struct todo_item));\n ++\tctx.items[ctx.items_nr].command = TODO_UPDATE_REFS;\n ++\tctx.items_nr++;\n ++\n +\tfree(todo_list->items);\n -+\ttodo_list->items = ctx.new_list.items;\n -+\ttodo_list->nr = ctx.new_list.nr;\n -+\ttodo_list->alloc = ctx.new_list.alloc;\n ++\ttodo_list->items = ctx.items;\n ++\ttodo_list->nr = ctx.items_nr;\n ++\ttodo_list->alloc = ctx.items_alloc;\n +\n +\treturn 0;\n +}\n @@ sequencer.c: static int skip_unnecessary_picks(struct repository *r,\n \t\t const char *shortrevisions, const char *onto_name,\n \t\t struct commit *onto, const struct object_id *orig_head,\n \t\t struct string_list *commands, unsigned autosquash,\n -+\t\t unsigned keep_decorations,\n ++\t\t unsigned update_refs,\n \t\t struct todo_list *todo_list)\n {\n \tchar shortonto[GIT_MAX_HEXSZ + 1];\n @@ sequencer.c: int complete_action(struct repository *r, struct replay_opts *opts, unsigned fla\n - \tif (autosquash && todo_list_rearrange_squash(todo_list))\n - \t\treturn -1;\n + \t\titem->arg_len = item->arg_offset = item->flags = item->offset_in_buf = 0;\n + \t}\n \n -+\tif (keep_decorations && todo_list_add_branch_commands(todo_list))\n ++\tif (update_refs && todo_list_add_update_ref_commands(todo_list))\n +\t\treturn -1;\n +\n - \tif (commands->nr)\n - \t\ttodo_list_add_exec_commands(todo_list, commands);\n + \tif (autosquash && todo_list_rearrange_squash(todo_list))\n + \t\treturn -1;\n \n \n ## sequencer.h ##\n @@ sequencer.h: int complete_action(struct repository *r, struct replay_opts *opts,\n \t\t const char *shortrevisions, const char *onto_name,\n \t\t struct commit *onto, const struct object_id *orig_head,\n \t\t struct string_list *commands, unsigned autosquash,\n -+\t\t unsigned keep_decorations,\n ++\t\t unsigned update_refs,\n \t\t struct todo_list *todo_list);\n int todo_list_rearrange_squash(struct todo_list *todo_list);\n \n @@ t/t3404-rebase-interactive.sh: test_expect_success 'ORIG_HEAD is updated correct\n \ttest_cmp_rev ORIG_HEAD test-orig-head@{1}\n '\n \n -+test_expect_success '--update-refs adds git branch commands' '\n ++test_expect_success '--update-refs adds label and update-ref commands' '\n +\tgit checkout -b update-refs no-conflict-branch &&\n -+\ttest_commit extra fileX &&\n -+\tgit commit --amend --fixup=L &&\n ++\tgit branch -f base HEAD~4 &&\n ++\tgit branch -f first HEAD~3 &&\n ++\tgit branch -f second HEAD~3 &&\n ++\tgit branch -f third HEAD~1 &&\n ++\tgit commit --allow-empty --fixup=third &&\n ++\tgit branch -f shared-tip &&\n +\t(\n +\t\tset_cat_todo_editor &&\n -+\t\tgit branch -f base A &&\n -+\t\tgit branch -f first J &&\n -+\t\tgit branch -f second J &&\n -+\t\tgit branch -f third L &&\n -+\n -+\t\ttest_must_fail git rebase -i --autosquash --update-refs primary >todo &&\n +\n +\t\tcat >expect <<-EOF &&\n +\t\tpick $(git log -1 --format=%h J) J\n -+\t\texec git update-ref refs/heads/second HEAD $(git rev-parse J)\n -+\t\texec git update-ref refs/heads/first HEAD $(git rev-parse J)\n ++\t\tlabel for-update-refs/refs/heads/second\n ++\t\tlabel for-update-refs/refs/heads/first\n +\t\tpick $(git log -1 --format=%h K) K\n +\t\tpick $(git log -1 --format=%h L) L\n -+\t\tfixup $(git log -1 --format=%h update-refs) fixup! L\n -+\t\texec git update-ref refs/heads/third HEAD $(git rev-parse L)\n ++\t\tfixup $(git log -1 --format=%h update-refs) fixup! L # empty\n ++\t\tlabel for-update-refs/refs/heads/third\n +\t\tpick $(git log -1 --format=%h M) M\n -+\t\texec git update-ref refs/heads/no-conflict-branch HEAD $(git rev-parse M)\n ++\t\tlabel for-update-refs/refs/heads/no-conflict-branch\n ++\t\tlabel for-update-refs/refs/heads/shared-tip\n ++\t\tupdate-refs\n +\t\tEOF\n +\n ++\t\ttest_must_fail git rebase -i --autosquash --update-refs primary >todo &&\n ++\t\ttest_cmp expect todo\n ++\t)\n ++'\n ++\n ++test_expect_success '--update-refs adds commands with --rebase-merges' '\n ++\tgit checkout -b update-refs-with-merge no-conflict-branch &&\n ++\tgit branch -f base HEAD~4 &&\n ++\tgit branch -f first HEAD~3 &&\n ++\tgit branch -f second HEAD~3 &&\n ++\tgit branch -f third HEAD~1 &&\n ++\tgit merge -m merge branch2 &&\n ++\tgit branch -f merge-branch &&\n ++\tgit commit --fixup=third --allow-empty &&\n ++\t(\n ++\t\tset_cat_todo_editor &&\n ++\n ++\t\tcat >expect <<-EOF &&\n ++\t\tlabel onto\n ++\t\treset onto\n ++\t\tpick $(git log -1 --format=%h branch2~1) F\n ++\t\tpick $(git log -1 --format=%h branch2) I\n ++\t\tlabel for-update-refs/refs/heads/branch2\n ++\t\tlabel merge\n ++\t\treset onto\n ++\t\tpick $(git log -1 --format=%h refs/heads/second) J\n ++\t\tlabel for-update-refs/refs/heads/second\n ++\t\tlabel for-update-refs/refs/heads/first\n ++\t\tpick $(git log -1 --format=%h refs/heads/third~1) K\n ++\t\tpick $(git log -1 --format=%h refs/heads/third) L\n ++\t\tfixup $(git log -1 --format=%h update-refs-with-merge) fixup! L # empty\n ++\t\tlabel for-update-refs/refs/heads/third\n ++\t\tpick $(git log -1 --format=%h HEAD~2) M\n ++\t\tlabel for-update-refs/refs/heads/no-conflict-branch\n ++\t\tmerge -C $(git log -1 --format=%h HEAD~1) merge # merge\n ++\t\tlabel for-update-refs/refs/heads/merge-branch\n ++\t\tupdate-refs\n ++\t\tEOF\n ++\n ++\t\ttest_must_fail git rebase -i --autosquash \\\n ++\t\t\t\t --rebase-merges=rebase-cousins \\\n ++\t\t\t\t --update-refs primary >todo &&\n ++\n +\t\ttest_cmp expect todo\n +\t)\n +'\n -: ----------- > 6: 68f8e51b19c sequencer: implement 'update-refs' command\n 4: b99c5bf34ef ! 7: 3d7d3f656b4 rebase: add rebase.updateRefs config option\n @@ builtin/rebase.c: static int rebase_config(const char *var, const char *value, v\n \t\treturn 0;\n \n ## t/t3404-rebase-interactive.sh ##\n -@@ t/t3404-rebase-interactive.sh: test_expect_success '--update-refs adds git branch commands' '\n - \t\texec git update-ref refs/heads/no-conflict-branch HEAD $(git rev-parse M)\n +@@ t/t3404-rebase-interactive.sh: test_expect_success '--update-refs adds label and update-ref commands' '\n \t\tEOF\n \n + \t\ttest_must_fail git rebase -i --autosquash --update-refs primary >todo &&\n +\t\ttest_cmp expect todo &&\n +\n +\t\ttest_must_fail git -c rebase.autosquash=true \\\n +\t\t\t\t -c rebase.updaterefs=true \\\n +\t\t\t\t rebase -i primary >todo &&\n ++\n + \t\ttest_cmp expect todo\n + \t)\n + '\n +@@ t/t3404-rebase-interactive.sh: test_expect_success '--update-refs adds commands with --rebase-merges' '\n + \t\t\t\t --rebase-merges=rebase-cousins \\\n + \t\t\t\t --update-refs primary >todo &&\n + \n ++\t\ttest_cmp expect todo &&\n ++\n ++\t\ttest_must_fail git -c rebase.autosquash=true \\\n ++\t\t\t\t -c rebase.updaterefs=true \\\n ++\t\t\t\t rebase -i \\\n ++\t\t\t\t --rebase-merges=rebase-cousins \\\n ++\t\t\t\t primary >todo &&\n ++\n \t\ttest_cmp expect todo\n \t)\n '\n\n-- \ngitgitgadget\n"},{"id":"456830","messageId":"4f9f34876413927d819313a70fcdefcad5b35689.1654634569.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","subject":"[PATCH v2 1/7] log-tree: create for_each_decoration()","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-07T20:42:43Z","receivedAt":"2022-06-08T00:33:40Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nIt can be helpful to iterate through all the decorations on a commit\nwithout necessarily writing them to a stream. Implement\nfor_each_decoration() and reimplement format_decorations_extended() to\nuse that iterator.\n\nSigned-off-by: Derrick Stolee \n---\n log-tree.c | 111 ++++++++++++++++++++++++++++++++++++-----------------\n log-tree.h | 4 ++\n 2 files changed, 80 insertions(+), 35 deletions(-)\n\ndiff --git a/log-tree.c b/log-tree.c\nindex d0ac0a6327a..b15a7c9db22 100644\n--- a/log-tree.c\n+++ b/log-tree.c\n@@ -282,6 +282,54 @@ static void show_name(struct strbuf *sb, const struct name_decoration *decoratio\n \t\tstrbuf_addstr(sb, decoration->name);\n }\n \n+struct format_decorations_context {\n+\tstruct strbuf *sb;\n+\tint use_color;\n+\tconst char *prefix;\n+\tconst char *separator;\n+\tconst char *suffix;\n+\tconst char *color_commit;\n+\tconst char *color_reset;\n+\tconst struct name_decoration *current_and_HEAD;\n+};\n+\n+static int append_decoration(const struct name_decoration *d,\n+\t\t\t void *data)\n+{\n+\tstruct format_decorations_context *ctx = data;\n+\t/*\n+\t * When both current and HEAD are there, only\n+\t * show HEAD->current where HEAD would have\n+\t * appeared, skipping the entry for current.\n+\t */\n+\tif (d != ctx->current_and_HEAD) {\n+\t\tstrbuf_addstr(ctx->sb, ctx->color_commit);\n+\t\tstrbuf_addstr(ctx->sb, ctx->prefix);\n+\t\tstrbuf_addstr(ctx->sb, ctx->color_reset);\n+\t\tstrbuf_addstr(ctx->sb, decorate_get_color(ctx->use_color, d->type));\n+\t\tif (d->type == DECORATION_REF_TAG)\n+\t\t\tstrbuf_addstr(ctx->sb, \"tag: \");\n+\n+\t\tshow_name(ctx->sb, d);\n+\n+\t\tif (ctx->current_and_HEAD &&\n+\t\t d->type == DECORATION_REF_HEAD) {\n+\t\t\tstrbuf_addstr(ctx->sb, \" -> \");\n+\t\t\tstrbuf_addstr(ctx->sb, ctx->color_reset);\n+\t\t\tstrbuf_addstr(ctx->sb,\n+\t\t\t\t decorate_get_color(\n+\t\t\t\t\t\tctx->use_color,\n+\t\t\t\t\t\tctx->current_and_HEAD->type));\n+\t\t\tshow_name(ctx->sb, ctx->current_and_HEAD);\n+\t\t}\n+\t\tstrbuf_addstr(ctx->sb, ctx->color_reset);\n+\n+\t\tctx->prefix = ctx->separator;\n+\t}\n+\n+\treturn 0;\n+}\n+\n /*\n * The caller makes sure there is no funny color before calling.\n * format_decorations_extended makes sure the same after return.\n@@ -294,49 +342,42 @@ void format_decorations_extended(struct strbuf *sb,\n \t\t\tconst char *suffix)\n {\n \tconst struct name_decoration *decoration;\n-\tconst struct name_decoration *current_and_HEAD;\n-\tconst char *color_commit =\n-\t\tdiff_get_color(use_color, DIFF_COMMIT);\n-\tconst char *color_reset =\n-\t\tdecorate_get_color(use_color, DECORATION_NONE);\n+\tstruct format_decorations_context ctx = {\n+\t\t.sb = sb,\n+\t\t.use_color = use_color,\n+\t\t.prefix = prefix,\n+\t\t.separator = separator,\n+\t\t.suffix = suffix,\n+\t\t.color_commit =\tdiff_get_color(use_color, DIFF_COMMIT),\n+\t\t.color_reset = decorate_get_color(use_color, DECORATION_NONE),\n+\t};\n \n \tdecoration = get_name_decoration(&commit->object);\n \tif (!decoration)\n \t\treturn;\n \n-\tcurrent_and_HEAD = current_pointed_by_HEAD(decoration);\n-\twhile (decoration) {\n-\t\t/*\n-\t\t * When both current and HEAD are there, only\n-\t\t * show HEAD->current where HEAD would have\n-\t\t * appeared, skipping the entry for current.\n-\t\t */\n-\t\tif (decoration != current_and_HEAD) {\n-\t\t\tstrbuf_addstr(sb, color_commit);\n-\t\t\tstrbuf_addstr(sb, prefix);\n-\t\t\tstrbuf_addstr(sb, color_reset);\n-\t\t\tstrbuf_addstr(sb, decorate_get_color(use_color, decoration->type));\n-\t\t\tif (decoration->type == DECORATION_REF_TAG)\n-\t\t\t\tstrbuf_addstr(sb, \"tag: \");\n-\n-\t\t\tshow_name(sb, decoration);\n-\n-\t\t\tif (current_and_HEAD &&\n-\t\t\t decoration->type == DECORATION_REF_HEAD) {\n-\t\t\t\tstrbuf_addstr(sb, \" -> \");\n-\t\t\t\tstrbuf_addstr(sb, color_reset);\n-\t\t\t\tstrbuf_addstr(sb, decorate_get_color(use_color, current_and_HEAD->type));\n-\t\t\t\tshow_name(sb, current_and_HEAD);\n-\t\t\t}\n-\t\t\tstrbuf_addstr(sb, color_reset);\n+\tctx.current_and_HEAD = current_pointed_by_HEAD(decoration);\n \n-\t\t\tprefix = separator;\n-\t\t}\n+\tfor_each_decoration(commit, append_decoration, &ctx);\n+\n+\tstrbuf_addstr(sb, ctx.color_commit);\n+\tstrbuf_addstr(sb, ctx.suffix);\n+\tstrbuf_addstr(sb, ctx.color_reset);\n+}\n+\n+int for_each_decoration(const struct commit *c, decoration_fn fn, void *data)\n+{\n+\tconst struct name_decoration *decoration;\n+\n+\tdecoration = get_name_decoration(&c->object);\n+\twhile (decoration) {\n+\t\tint res;\n+\t\tif ((res = fn(decoration, data)))\n+\t\t\treturn res;\n \t\tdecoration = decoration->next;\n \t}\n-\tstrbuf_addstr(sb, color_commit);\n-\tstrbuf_addstr(sb, suffix);\n-\tstrbuf_addstr(sb, color_reset);\n+\n+\treturn 0;\n }\n \n void show_decorations(struct rev_info *opt, struct commit *commit)\ndiff --git a/log-tree.h b/log-tree.h\nindex e7e4641cf83..ea07da2625b 100644\n--- a/log-tree.h\n+++ b/log-tree.h\n@@ -35,4 +35,8 @@ void fmt_output_commit(struct strbuf *, struct commit *, struct rev_info *);\n void fmt_output_subject(struct strbuf *, const char *subject, struct rev_info *);\n void fmt_output_email_subject(struct strbuf *, struct rev_info *);\n \n+typedef int decoration_fn(const struct name_decoration *d,\n+\t\t\t void *data);\n+int for_each_decoration(const struct commit *c, decoration_fn fn, void *data);\n+\n #endif\n-- \ngitgitgadget\n\n"},{"id":"456836","messageId":"xmqqmteot98j.fsf@gitster.g","threadId":"57949","inReplyTo":"9f261c7df2c0a433c6acc46ffefd8d2633fdc3d4.1654634569.git.gitgitgadget@gmail.com","subject":"Re: [PATCH v2 3/7] sequencer: define array with enum values","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-07T22:11:08Z","receivedAt":"2022-06-08T00:33:50Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"\"Derrick Stolee via GitGitGadget\" writes:\n\n> From: Derrick Stolee \n>\n> The todo_command_info array defines which strings match with which\n> todo_command enum values. The array is defined in the same order as the\n> enum values, but if one changed without the other, then we would have\n> unexpected results.\n>\n> Make it easier to see changes to the enum and this array by using the\n> enum values as the indices of the array.\n>\n> Signed-off-by: Derrick Stolee \n> ---\n> sequencer.c | 28 ++++++++++++++--------------\n> 1 file changed, 14 insertions(+), 14 deletions(-)\n\nGood thinking.\n\nI love seeing a future-proofing change like this.\n\nThanks.\n\n> diff --git a/sequencer.c b/sequencer.c\n> index 8c3ed3532ac..8e26c9a6261 100644\n> --- a/sequencer.c\n> +++ b/sequencer.c\n> @@ -1685,20 +1685,20 @@ static struct {\n> \tchar c;\n> \tconst char *str;\n> } todo_command_info[] = {\n> -\t{ 'p', \"pick\" },\n> -\t{ 0, \"revert\" },\n> -\t{ 'e', \"edit\" },\n> -\t{ 'r', \"reword\" },\n> -\t{ 'f', \"fixup\" },\n> -\t{ 's', \"squash\" },\n> -\t{ 'x', \"exec\" },\n> -\t{ 'b', \"break\" },\n> -\t{ 'l', \"label\" },\n> -\t{ 't', \"reset\" },\n> -\t{ 'm', \"merge\" },\n> -\t{ 0, \"noop\" },\n> -\t{ 'd', \"drop\" },\n> -\t{ 0, NULL }\n> +\t[TODO_PICK] = { 'p', \"pick\" },\n> +\t[TODO_REVERT] = { 0, \"revert\" },\n> +\t[TODO_EDIT] = { 'e', \"edit\" },\n> +\t[TODO_REWORD] = { 'r', \"reword\" },\n> +\t[TODO_FIXUP] = { 'f', \"fixup\" },\n> +\t[TODO_SQUASH] = { 's', \"squash\" },\n> +\t[TODO_EXEC] = { 'x', \"exec\" },\n> +\t[TODO_BREAK] = { 'b', \"break\" },\n> +\t[TODO_LABEL] = { 'l', \"label\" },\n> +\t[TODO_RESET] = { 't', \"reset\" },\n> +\t[TODO_MERGE] = { 'm', \"merge\" },\n> +\t[TODO_NOOP] = { 0, \"noop\" },\n> +\t[TODO_DROP] = { 'd', \"drop\" },\n> +\t[TODO_COMMENT] = { 0, NULL },\n> };\n> \n> static const char *command_to_string(const enum todo_command command)\n"},{"id":"456837","messageId":"9f261c7df2c0a433c6acc46ffefd8d2633fdc3d4.1654634569.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","subject":"[PATCH v2 3/7] sequencer: define array with enum values","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-07T20:42:45Z","receivedAt":"2022-06-08T01:20:39Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe todo_command_info array defines which strings match with which\ntodo_command enum values. The array is defined in the same order as the\nenum values, but if one changed without the other, then we would have\nunexpected results.\n\nMake it easier to see changes to the enum and this array by using the\nenum values as the indices of the array.\n\nSigned-off-by: Derrick Stolee \n---\n sequencer.c | 28 ++++++++++++++--------------\n 1 file changed, 14 insertions(+), 14 deletions(-)\n\ndiff --git a/sequencer.c b/sequencer.c\nindex 8c3ed3532ac..8e26c9a6261 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -1685,20 +1685,20 @@ static struct {\n \tchar c;\n \tconst char *str;\n } todo_command_info[] = {\n-\t{ 'p', \"pick\" },\n-\t{ 0, \"revert\" },\n-\t{ 'e', \"edit\" },\n-\t{ 'r', \"reword\" },\n-\t{ 'f', \"fixup\" },\n-\t{ 's', \"squash\" },\n-\t{ 'x', \"exec\" },\n-\t{ 'b', \"break\" },\n-\t{ 'l', \"label\" },\n-\t{ 't', \"reset\" },\n-\t{ 'm', \"merge\" },\n-\t{ 0, \"noop\" },\n-\t{ 'd', \"drop\" },\n-\t{ 0, NULL }\n+\t[TODO_PICK] = { 'p', \"pick\" },\n+\t[TODO_REVERT] = { 0, \"revert\" },\n+\t[TODO_EDIT] = { 'e', \"edit\" },\n+\t[TODO_REWORD] = { 'r', \"reword\" },\n+\t[TODO_FIXUP] = { 'f', \"fixup\" },\n+\t[TODO_SQUASH] = { 's', \"squash\" },\n+\t[TODO_EXEC] = { 'x', \"exec\" },\n+\t[TODO_BREAK] = { 'b', \"break\" },\n+\t[TODO_LABEL] = { 'l', \"label\" },\n+\t[TODO_RESET] = { 't', \"reset\" },\n+\t[TODO_MERGE] = { 'm', \"merge\" },\n+\t[TODO_NOOP] = { 0, \"noop\" },\n+\t[TODO_DROP] = { 'd', \"drop\" },\n+\t[TODO_COMMENT] = { 0, NULL },\n };\n \n static const char *command_to_string(const enum todo_command command)\n-- \ngitgitgadget\n\n"},{"id":"456838","messageId":"5f54766e1032ebf3a331516a6dd696b997bdfdd8.1654634569.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","subject":"[PATCH v2 2/7] branch: add branch_checked_out() helper","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-07T20:42:44Z","receivedAt":"2022-06-08T01:20:45Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe validate_new_branchname() method contains a check to see if a branch\nis checked out in any non-bare worktree. This is intended to prevent a\nforce push that will mess up an existing checkout. This helper is not\nsuitable to performing just that check, because the method will die()\nwhen the branch is checked out instead of returning an error code.\n\nExtract branch_checked_out() and use it within\nvalidate_new_branchname(). Another caller will be added in a coming\nchange.\n\nSigned-off-by: Derrick Stolee \n---\n branch.c | 24 ++++++++++++++++--------\n branch.h | 8 ++++++++\n 2 files changed, 24 insertions(+), 8 deletions(-)\n\ndiff --git a/branch.c b/branch.c\nindex 2d6569b0c62..2e6419cdfa5 100644\n--- a/branch.c\n+++ b/branch.c\n@@ -369,6 +369,19 @@ int validate_branchname(const char *name, struct strbuf *ref)\n \treturn ref_exists(ref->buf);\n }\n \n+int branch_checked_out(const char *refname, char **path)\n+{\n+\tstruct worktree **worktrees = get_worktrees();\n+\tconst struct worktree *wt = find_shared_symref(worktrees, \"HEAD\", refname);\n+\tint result = wt && !wt->is_bare;\n+\n+\tif (result && path)\n+\t\t*path = xstrdup(wt->path);\n+\n+\tfree_worktrees(worktrees);\n+\treturn result;\n+}\n+\n /*\n * Check if a branch 'name' can be created as a new branch; die otherwise.\n * 'force' can be used when it is OK for the named branch already exists.\n@@ -377,9 +390,7 @@ int validate_branchname(const char *name, struct strbuf *ref)\n */\n int validate_new_branchname(const char *name, struct strbuf *ref, int force)\n {\n-\tstruct worktree **worktrees;\n-\tconst struct worktree *wt;\n-\n+\tchar *path;\n \tif (!validate_branchname(name, ref))\n \t\treturn 0;\n \n@@ -387,13 +398,10 @@ int validate_new_branchname(const char *name, struct strbuf *ref, int force)\n \t\tdie(_(\"a branch named '%s' already exists\"),\n \t\t ref->buf + strlen(\"refs/heads/\"));\n \n-\tworktrees = get_worktrees();\n-\twt = find_shared_symref(worktrees, \"HEAD\", ref->buf);\n-\tif (wt && !wt->is_bare)\n+\tif (branch_checked_out(ref->buf, &path))\n \t\tdie(_(\"cannot force update the branch '%s' \"\n \t\t \"checked out at '%s'\"),\n-\t\t ref->buf + strlen(\"refs/heads/\"), wt->path);\n-\tfree_worktrees(worktrees);\n+\t\t ref->buf + strlen(\"refs/heads/\"), path);\n \n \treturn 1;\n }\ndiff --git a/branch.h b/branch.h\nindex 560b6b96a8f..5ea93d217b1 100644\n--- a/branch.h\n+++ b/branch.h\n@@ -101,6 +101,14 @@ void create_branches_recursively(struct repository *r, const char *name,\n \t\t\t\t const char *tracking_name, int force,\n \t\t\t\t int reflog, int quiet, enum branch_track track,\n \t\t\t\t int dry_run);\n+\n+/*\n+ * Returns true if the branch at 'refname' is checked out at any\n+ * non-bare worktree. The path of the worktree is stored in the\n+ * given 'path', if provided.\n+ */\n+int branch_checked_out(const char *refname, char **path);\n+\n /*\n * Check if 'name' can be a valid name for a branch; die otherwise.\n * Return 1 if the named branch already exists; return 0 otherwise.\n-- \ngitgitgadget\n\n"},{"id":"456841","messageId":"9354d1d3-c1b7-3baf-215f-30659ad48b22@github.com","threadId":"57949","inReplyTo":"90ccd923-3552-fe88-5d6d-869def7f1aeb@gmail.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Derrick Stolee","fromEmail":"derrickstolee@github.com","sentAt":"2022-06-07T19:39:37Z","receivedAt":"2022-06-08T01:31:07Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"On 6/7/2022 6:11 AM, Phillip Wood wrote:\n> On 06/06/2022 16:12, Derrick Stolee wrote:\n>> On 6/4/2022 11:28 AM, Phillip Wood wrote:\n>>> On 03/06/2022 19:27, Taylor Blau wrote:\n>>>> On Fri, Jun 03, 2022 at 01:37:48PM +0000, Derrick Stolee via GitGitGadget wrote:\n>>>>> This is a feature I've wanted for quite a while. When working on the sparse\n>>>>> index topic, I created a long RFC that actually broke into three topics for\n>>>>> full review upstream. These topics were sequential, so any feedback on an\n>>>>> earlier one required updates to the later ones. I would work on the full\n>>>>> feature and use interactive rebase to update the full list of commits.\n>>>>> However, I would need to update the branches pointing to those sub-topics.\n>>>>\n>>>> This is really exciting. I'm glad that you're working on it, because I\n>>>> have also wanted this feature a handful of times over the years.\n>>>\n>>> Yes, thank you Stolee. I agree this will be useful, but I wonder if users\n>>> will be confused that --update-refs only updates the branch heads that happen\n>>> to be in the todo list, rather than updating all the branches that contain a\n>>> rewritten commit. I think the latter is something we should try to add in the\n>>> future and so we should take care to design this topic so that is possible.\n>>\n>> At the moment, the design adds a comment to the TODO list, showing which\n>> branches are not possible to move because they are checked out at another\n>> worktree (or is currently checked out and will be updated by the rebase\n>> itself). That seems like a good place to insert alternative logic in the\n>> future if we see a need for better behavior here.\n> \n> I think the question of whether to update branches that are checked out in\n> another worktree is a question of whether it is less inconvenient to the user\n> to skip the ref update and leave the user to manually update the branch or to> update the ref and leave the worktree in a potentially awkward state if the\n> user was half way through building a commit. The answer probably depends on\n> the preferences of the user.\n\nI think that their 'git status' will look strange no matter what: their\nworking directory and index could look significantly different from what\nthe branch at HEAD is reporting. For this situation, I would rather continue\npreventing these ref updates from underneath worktrees.\n \n> I've been using a script that updates the refs for all the branches being\n> rewritten for a while and have found it preferable to always update the ref\n> rather than have to do it manually. My script also updates the worktree\n> checkout to the new HEAD if there are no uncommitted changes which I have\n> found very convenient. My preference is probably because I tend not to have\n> uncommitted changes lying around in the worktrees whose branches get updated.\n\nActually updating the worktree to match seems like an interesting twist, which\nwe would want to consider if we go this route in the future.\n \n>> Unless: am I misunderstanding something about your concern here? Are you\n>> worried about refs outside of refs/heads/*? Are you concerned about it\n>> being _all_ refs/heads/* that we find?\n> \n> My concerns are primarily about being able to extend the --update-ref\n> option in a backwards compatible way.\n> \n> I'd like to be able to add functionality to rebase all refs/heads/* that\n> are descended from the commits that a simple rebase would rewrite. Say I want\n> to edit $commit then I want the rebase to rewrite all the commits and update\n> all the branches in in\n> \n>     git rev-list $(git for-each-ref --contains $commit refs/heads/) ^$commit^@\n> \n> Ideally we'd avoid adding a new commandline option when adding that. I think\n> we could use an optional argument as you suggest below (though it would not be\n> a refspec).\n\nThis is sort of the opposite of what my series is doing: you want to find all\nrefs that contain the current commits and make sure that they are _also_\nrebased as we rebase the current set of commits. That seems like an interesting\ndirection, with a very different set of complexities (ignoring other worktrees\nseems like much more of a blocker here).\n\nI would consider this to be _yet another mode_ that is separate from\n--update-refs, though it could share some underlying logic. It gets more\ncomplicated if there are merge commits involved (do we have a --rebase-merges\noption?).\n\nThis makes me think of earlier discussions around a \"git evolve\" command\n(based on Mercurial's evolve command). The idea is to have a different\nworkflow than I am presenting: instead of creating `fixup!` commits at the\ntop of a multi-part topic, you could check out a branch on a middle part\nand work forward from there. Then, you could \"rebase every branch with a\nrewritten commit\" to get things back into a nice line.\n\nIn conclusion: I don't think we should go down this rabbit hole right now.\nI think that --update-refs would be something we want to enable by default\nif we have such a mode, though.\n \n>> One potential way to extend this (in the future) is to make --update-refs\n>> take an optional string argument containing a refspec. This would replace\n>> the default refspec of refs/heads/*.\n>> [...] \n>>> Instead of using 'label' and 'exec' I'd prefer a new todo list command\n>>> ('update-ref' or 'update-branch'?) used in place of 'label' that takes a\n>>> branch name and updates the branch ref at the end of the rebase. That\n>>> would make it easy to do all the updates in a single transaction as you\n>>> suggested. Adding exec lines to do this makes the todo list messy and we\n>>> have been trying to stop rebase forking all the time.\n>>\n>> Thanks. Yes, a new 'update-refs' step at the end would be good to make\n>> it clear we want to rewrite the refs in one go without a possible\n>> interruption from the user.\n> \n> That's great, it is a bit more work for you but I think it gives a much\n> nicer UI.\n\nThanks. After some hurdles on my end (and some additional complexities\ndiscovered in the process) I will have v2 ready soon.\n\nThanks,\n-Stolee\n"},{"id":"456842","messageId":"xmqqr140t9am.fsf@gitster.g","threadId":"57949","inReplyTo":"5f54766e1032ebf3a331516a6dd696b997bdfdd8.1654634569.git.gitgitgadget@gmail.com","subject":"Re: [PATCH v2 2/7] branch: add branch_checked_out() helper","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-07T22:09:53Z","receivedAt":"2022-06-08T01:42:15Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"\"Derrick Stolee via GitGitGadget\" writes:\n\n> +int branch_checked_out(const char *refname, char **path)\n> +{\n> +\tstruct worktree **worktrees = get_worktrees();\n> +\tconst struct worktree *wt = find_shared_symref(worktrees, \"HEAD\", refname);\n> +\tint result = wt && !wt->is_bare;\n> +\n> +\tif (result && path)\n> +\t\t*path = xstrdup(wt->path);\n> +\n> +\tfree_worktrees(worktrees);\n> +\treturn result;\n> +}\n\nDon't you plan to call this repeatedly from the for_each_deco\niteration? I am wondering if it should take the result of\nget_worktrees() and reuse the result of get_worktrees(), instead of\nenumerating the list of worktrees and freeing for each of the\nbranches you need to inspect.\n\nThere also was another topic\n\nhttps://lore.kernel.org/git/pull.1266.v2.git.git.1652690501963.gitgitgadget@gmail.com/\n\nthat was triggered by find_shared_symref() being relatively\nheavy-weight, which suggests a more involved refactoring.\n\nI wonder if we rather want to rewrite find_shared_symref() *not* to\ntake the target parameter at all, and instead introduce a new\nfunction that takes a worktree, and report the branch that is\nchecked out (or being operated on via rebase or bisect). Then we\ncan\n\n - create a strset out of its result, i.e. set of branches that\n should not be touched;\n\n - iterate over refs that point into the history being rebased\n (using for_each_decoration()), and consult that strset to see if\n any of them is being rewritten.\n\nWith the API of find_shared_symref(), we'd need to iterate over all\nworktrees for each decoration. With such a restructuring, we can\niterate over all worktrees just once, and match the result with\ndecoration, so the problem becomes O(N)+O(M) and not O(N*M) for\nnumber of worktrees N and number of decorations M.\n"},{"id":"456843","messageId":"xmqqilpct8oj.fsf@gitster.g","threadId":"57949","inReplyTo":"68f8e51b19c024e1cb075dffd07f81cceaba5957.1654634569.git.gitgitgadget@gmail.com","subject":"Re: [PATCH v2 6/7] sequencer: implement 'update-refs' command","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-07T22:23:08Z","receivedAt":"2022-06-08T02:15:59Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"\"Derrick Stolee via GitGitGadget\" writes:\n\n> From: Derrick Stolee \n>\n> The previous change allowed 'git rebase --update-refs' to create 'label'\n> commands for each branch among the commits being rewritten and add an\n> 'update-refs' command at the end of the todo list. Now, teach Git to\n> update the refs during that final 'update-refs' command.\n>\n> We need to create an array of new and old OIDs for each ref by iterating\n> over the refs/rewritten/for-update-refs/ namespace. We cannot update the\n> refs in-place since this will confuse the refs iterator.\n\nIn other words, grab everything we need to do and then use that\nin-core information to tell refs API what refs to change to what\nvalues? That dounds like a quite reasonable thing to do.\n\nLooking at the patch text, the only thing that stands out at me is\nthat this does not seem to perform the updates in a single\ntransaction, which may often not matter in practice, but may be a\nprudent thing to do anyway at philosophical level.\n\nThanks, will queue.\n"},{"id":"456845","messageId":"842b2186d250aca367e9680105d4d15f8cbb098e.1654634569.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","subject":"[PATCH v2 4/7] sequencer: add update-refs command","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-07T20:42:46Z","receivedAt":"2022-06-08T02:16:17Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nAdd the boilerplat for an \"update-refs\" command in the sequencer. This\nconnects to the current no-op do_update_refs() which will be filled in\nafter more connections are created.\n\nSigned-off-by: Derrick Stolee \n---\n sequencer.c | 13 ++++++++++++-\n sequencer.h | 1 +\n 2 files changed, 13 insertions(+), 1 deletion(-)\n\ndiff --git a/sequencer.c b/sequencer.c\nindex 8e26c9a6261..68f7c76e896 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -1696,6 +1696,7 @@ static struct {\n \t[TODO_LABEL] = { 'l', \"label\" },\n \t[TODO_RESET] = { 't', \"reset\" },\n \t[TODO_MERGE] = { 'm', \"merge\" },\n+\t[TODO_UPDATE_REFS] = { 'u', \"update-refs\" },\n \t[TODO_NOOP] = { 0, \"noop\" },\n \t[TODO_DROP] = { 'd', \"drop\" },\n \t[TODO_COMMENT] = { 0, NULL },\n@@ -2442,7 +2443,9 @@ static int parse_insn_line(struct repository *r, struct todo_item *item,\n \tpadding = strspn(bol, \" \\t\");\n \tbol += padding;\n \n-\tif (item->command == TODO_NOOP || item->command == TODO_BREAK) {\n+\tif (item->command == TODO_NOOP ||\n+\t item->command == TODO_BREAK ||\n+\t item->command == TODO_UPDATE_REFS) {\n \t\tif (bol != eol)\n \t\t\treturn error(_(\"%s does not accept arguments: '%s'\"),\n \t\t\t\t command_to_string(item->command), bol);\n@@ -4056,6 +4059,11 @@ leave_merge:\n \treturn ret;\n }\n \n+static int do_update_refs(struct repository *r)\n+{\n+\treturn 0;\n+}\n+\n static int is_final_fixup(struct todo_list *todo_list)\n {\n \tint i = todo_list->current;\n@@ -4431,6 +4439,9 @@ static int pick_commits(struct repository *r,\n \t\t\t\treturn error_with_patch(r, item->commit,\n \t\t\t\t\t\t\targ, item->arg_len,\n \t\t\t\t\t\t\topts, res, 0);\n+\t\t} else if (item->command == TODO_UPDATE_REFS) {\n+\t\t\tif ((res = do_update_refs(r)))\n+\t\t\t\treschedule = 1;\n \t\t} else if (!is_noop(item->command))\n \t\t\treturn error(_(\"unknown command %d\"), item->command);\n \ndiff --git a/sequencer.h b/sequencer.h\nindex da64473636b..c2b4e148d8f 100644\n--- a/sequencer.h\n+++ b/sequencer.h\n@@ -95,6 +95,7 @@ enum todo_command {\n \tTODO_LABEL,\n \tTODO_RESET,\n \tTODO_MERGE,\n+\tTODO_UPDATE_REFS,\n \t/* commands that do nothing but are counted for reporting progress */\n \tTODO_NOOP,\n \tTODO_DROP,\n-- \ngitgitgadget\n\n"},{"id":"456846","messageId":"68f8e51b19c024e1cb075dffd07f81cceaba5957.1654634569.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","subject":"[PATCH v2 6/7] sequencer: implement 'update-refs' command","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-07T20:42:48Z","receivedAt":"2022-06-08T02:46:50Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe previous change allowed 'git rebase --update-refs' to create 'label'\ncommands for each branch among the commits being rewritten and add an\n'update-refs' command at the end of the todo list. Now, teach Git to\nupdate the refs during that final 'update-refs' command.\n\nWe need to create an array of new and old OIDs for each ref by iterating\nover the refs/rewritten/for-update-refs/ namespace. We cannot update the\nrefs in-place since this will confuse the refs iterator.\n\nSigned-off-by: Derrick Stolee \n---\n sequencer.c | 62 ++++++++++++++++++++++++++++++++++-\n t/t3404-rebase-interactive.sh | 24 ++++++++++++++\n 2 files changed, 85 insertions(+), 1 deletion(-)\n\ndiff --git a/sequencer.c b/sequencer.c\nindex 94f8d52e041..a8f62ce8e5f 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -4061,11 +4061,71 @@ leave_merge:\n \treturn ret;\n }\n \n-static int do_update_refs(struct repository *r)\n+struct update_refs_context {\n+\tstruct ref_store *refs;\n+\tchar **ref_names;\n+\tstruct object_id *old;\n+\tstruct object_id *new;\n+\tsize_t nr;\n+\tsize_t alloc;\n+};\n+\n+static int add_ref_to_context(const char *refname,\n+\t\t\t const struct object_id *oid,\n+\t\t\t int flags,\n+\t\t\t void *data)\n {\n+\tint f = 0;\n+\tconst char *name;\n+\tstruct update_refs_context *ctx = data;\n+\n+\tALLOC_GROW(ctx->ref_names, ctx->nr + 1, ctx->alloc);\n+\tALLOC_GROW(ctx->old, ctx->nr + 1, ctx->alloc);\n+\tALLOC_GROW(ctx->new, ctx->nr + 1, ctx->alloc);\n+\n+\tif (!skip_prefix(refname, \"refs/rewritten/for-update-refs/\", &name))\n+\t\treturn 1;\n+\n+\tctx->ref_names[ctx->nr] = xstrdup(name);\n+\toidcpy(&ctx->new[ctx->nr], oid);\n+\tif (!refs_resolve_ref_unsafe(ctx->refs, name, 0,\n+\t\t\t\t &ctx->old[ctx->nr], &f))\n+\t\treturn 1;\n+\n+\tctx->nr++;\n \treturn 0;\n }\n \n+static int do_update_refs(struct repository *r)\n+{\n+\tint i, res;\n+\tstruct update_refs_context ctx = {\n+\t\t.refs = get_main_ref_store(r),\n+\t\t.alloc = 16,\n+\t};\n+\tALLOC_ARRAY(ctx.ref_names, ctx.alloc);\n+\tALLOC_ARRAY(ctx.old, ctx.alloc);\n+\tALLOC_ARRAY(ctx.new, ctx.alloc);\n+\n+\tres = refs_for_each_fullref_in(ctx.refs,\n+\t\t\t\t \"refs/rewritten/for-update-refs/\",\n+\t\t\t\t add_ref_to_context,\n+\t\t\t\t &ctx);\n+\n+\tfor (i = 0; !res && i < ctx.nr; i++)\n+\t\tres = refs_update_ref(ctx.refs, \"rewritten during rebase\",\n+\t\t\t\tctx.ref_names[i],\n+\t\t\t\t&ctx.new[i], &ctx.old[i],\n+\t\t\t\t0, UPDATE_REFS_MSG_ON_ERR);\n+\n+\tfor (i = 0; i < ctx.nr; i++)\n+\t\tfree(ctx.ref_names[i]);\n+\tfree(ctx.ref_names);\n+\tfree(ctx.old);\n+\tfree(ctx.new);\n+\treturn res;\n+}\n+\n static int is_final_fixup(struct todo_list *todo_list)\n {\n \tint i = todo_list->current;\ndiff --git a/t/t3404-rebase-interactive.sh b/t/t3404-rebase-interactive.sh\nindex 5e99ad7f3b6..72711efec28 100755\n--- a/t/t3404-rebase-interactive.sh\n+++ b/t/t3404-rebase-interactive.sh\n@@ -1815,6 +1815,30 @@ test_expect_success '--update-refs adds commands with --rebase-merges' '\n \t)\n '\n \n+compare_two_refs () {\n+\tgit rev-parse $1 >expect &&\n+\tgit rev-parse $2 >actual &&\n+\ttest_cmp expect actual\n+}\n+\n+test_expect_success '--update-refs updates refs correctly' '\n+\tgit checkout -B update-refs no-conflict-branch &&\n+\tgit branch -f base HEAD~4 &&\n+\tgit branch -f first HEAD~3 &&\n+\tgit branch -f second HEAD~3 &&\n+\tgit branch -f third HEAD~1 &&\n+\ttest_commit extra2 fileX &&\n+\tgit commit --amend --fixup=L &&\n+\t(\n+\t\tgit rebase -i --autosquash --update-refs primary &&\n+\n+\t\tcompare_two_refs HEAD~3 refs/heads/first &&\n+\t\tcompare_two_refs HEAD~3 refs/heads/second &&\n+\t\tcompare_two_refs HEAD~1 refs/heads/third &&\n+\t\tcompare_two_refs HEAD refs/heads/no-conflict-branch\n+\t)\n+'\n+\n # This must be the last test in this file\n test_expect_success '$EDITOR and friends are unchanged' '\n \ttest_editor_unchanged\n-- \ngitgitgadget\n\n"},{"id":"456847","messageId":"0a4c110127b7628d3bc5b04534a4ea67fe46ef80.1654634569.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","subject":"[PATCH v2 5/7] rebase: add --update-refs option","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-07T20:42:47Z","receivedAt":"2022-06-08T02:47:14Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nWhen working on a large feature, it can be helpful to break that feature\ninto multiple smaller parts that become reviewed in sequence. During\ndevelopment or during review, a change to one part of the feature could\naffect multiple of these parts. An interactive rebase can help adjust\nthe multi-part \"story\" of the branch.\n\nHowever, if there are branches tracking the different parts of the\nfeature, then rebasing the entire list of commits can create commits not\nreachable from those \"sub branches\". It can take a manual step to update\nthose branches.\n\nAdd a new --update-refs option to 'git rebase -i' that adds 'label\nfor-update-refs/*' steps to the todo file whenever a commit that is\nbeing rebased is decorated with that . At the very end, the\n'update-refs' step is added to update all of the branches referenced by\nthe 'label' steps. This allows the user to rebase a long list of commits\nin a multi-part feature and keep all of their pointers to those parts.\n\nNOTE: This change only introduce the --update-refs option and implements\nthe changes to the todo file. It does _not_ yet implement the action\ntaken by the 'update-refs' todo step, which will be implemented and\ntested in a later change.\n\nUse the new for_each_decoration() while iterating over the existing todo\nlist. Be sure to iterate after any squashing or fixups are placed.\nUpdate the branch only after those squashes and fixups are complete.\nThis allows a --fixup commit at the tip of the feature to apply\ncorrectly to the sub branch, even if it is fixing up the most-recent\ncommit in that part.\n\nOne potential problem here is that refs decorating commits that are\nalready marked as \"fixup!\" or \"squash!\" will not be included in this\nlist. Generally, the reordering of the \"fixup!\" and \"squash!\" is likely\nto change the relative order of these refs, so it is not recommended.\nThe workflow here is intended to allow these kinds of commits at the tip\nof the rebased branch while the other sub branches come along for the\nride without intervention.\n\nBe careful to not attempt updating any branch that is checked out. The\nmost common example is the branch being rebased is checked out and\ndecorates the tip commit. If the user is rebasing commits reachable from\na different branch that is checked out in a different worktree, then\nthey may be surprised to not see that ref update. However, it's probably\nbest to not optimize for this scenario and do the safest thing that will\nresult in a successful rebase. A comment is left in the TODO list that\nsignals that these refs are currently checked out.\n\nSigned-off-by: Derrick Stolee \n---\n Documentation/git-rebase.txt | 8 +++\n builtin/rebase.c | 5 ++\n sequencer.c | 105 ++++++++++++++++++++++++++++++++++\n sequencer.h | 1 +\n t/t3404-rebase-interactive.sh | 72 +++++++++++++++++++++++\n 5 files changed, 191 insertions(+)\n\ndiff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt\nindex 262fb01aec0..e7611b4089c 100644\n--- a/Documentation/git-rebase.txt\n+++ b/Documentation/git-rebase.txt\n@@ -609,6 +609,13 @@ provided. Otherwise an explicit `--no-reschedule-failed-exec` at the\n start would be overridden by the presence of\n `rebase.rescheduleFailedExec=true` configuration.\n \n+--update-refs::\n+--no-update-refs::\n+\tAutomatically force-update any branches that point to commits that\n+\tare being rebased. Any branches that are checked out in a worktree\n+\tor point to a `squash! ...` or `fixup! ...` commit are not updated\n+\tin this way.\n+\n INCOMPATIBLE OPTIONS\n --------------------\n \n@@ -632,6 +639,7 @@ are incompatible with the following options:\n * --empty=\n * --reapply-cherry-picks\n * --edit-todo\n+ * --update-refs\n * --root when used in combination with --onto\n \n In addition, the following pairs of options are incompatible:\ndiff --git a/builtin/rebase.c b/builtin/rebase.c\nindex 7ab50cda2ad..56d82a52106 100644\n--- a/builtin/rebase.c\n+++ b/builtin/rebase.c\n@@ -102,6 +102,7 @@ struct rebase_options {\n \tint reschedule_failed_exec;\n \tint reapply_cherry_picks;\n \tint fork_point;\n+\tint update_refs;\n };\n \n #define REBASE_OPTIONS_INIT {\t\t\t \t\\\n@@ -298,6 +299,7 @@ static int do_interactive_rebase(struct rebase_options *opts, unsigned flags)\n \t\tret = complete_action(the_repository, &replay, flags,\n \t\t\tshortrevisions, opts->onto_name, opts->onto,\n \t\t\t&opts->orig_head, &commands, opts->autosquash,\n+\t\t\topts->update_refs,\n \t\t\t&todo_list);\n \t}\n \n@@ -1124,6 +1126,9 @@ int cmd_rebase(int argc, const char **argv, const char *prefix)\n \t\tOPT_BOOL(0, \"autosquash\", &options.autosquash,\n \t\t\t N_(\"move commits that begin with \"\n \t\t\t \"squash!/fixup! under -i\")),\n+\t\tOPT_BOOL(0, \"update-refs\", &options.update_refs,\n+\t\t\t N_(\"update local refs that point to commits \"\n+\t\t\t \"that are being rebased\")),\n \t\t{ OPTION_STRING, 'S', \"gpg-sign\", &gpg_sign, N_(\"key-id\"),\n \t\t\tN_(\"GPG-sign commits\"),\n \t\t\tPARSE_OPT_OPTARG, NULL, (intptr_t) \"\" },\ndiff --git a/sequencer.c b/sequencer.c\nindex 68f7c76e896..94f8d52e041 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -35,6 +35,8 @@\n #include \"commit-reach.h\"\n #include \"rebase-interactive.h\"\n #include \"reset.h\"\n+#include \"branch.h\"\n+#include \"log-tree.h\"\n \n #define GIT_REFLOG_ACTION \"GIT_REFLOG_ACTION\"\n \n@@ -5614,10 +5616,110 @@ static int skip_unnecessary_picks(struct repository *r,\n \treturn 0;\n }\n \n+struct todo_add_branch_context {\n+\tstruct todo_item *items;\n+\tsize_t items_nr;\n+\tsize_t items_alloc;\n+\tstruct strbuf *buf;\n+\tstruct commit *commit;\n+};\n+\n+static int add_branch_for_decoration(const struct name_decoration *d, void *data)\n+{\n+\tstruct todo_add_branch_context *ctx = data;\n+\tsize_t base_offset = ctx->buf->len;\n+\tstruct todo_item *item;\n+\tchar *path;\n+\n+\tALLOC_GROW(ctx->items,\n+\t\t ctx->items_nr + 1,\n+\t\t ctx->items_alloc);\n+\titem = &ctx->items[ctx->items_nr];\n+\tmemset(item, 0, sizeof(*item));\n+\n+\t/* If the branch is checked out, then leave a comment instead. */\n+\tif (branch_checked_out(d->name, &path)) {\n+\t\titem->command = TODO_COMMENT;\n+\t\tstrbuf_addf(ctx->buf, \"# Ref %s checked out at '%s'\\n\",\n+\t\t\t d->name, path);\n+\t\tfree(path);\n+\t} else {\n+\t\titem->command = TODO_LABEL;\n+\t\tstrbuf_addf(ctx->buf, \"for-update-refs/%s\\n\", d->name);\n+\t}\n+\n+\titem->offset_in_buf = base_offset;\n+\titem->arg_offset = base_offset;\n+\titem->arg_len = ctx->buf->len - base_offset;\n+\tctx->items_nr++;\n+\n+\treturn 0;\n+}\n+\n+/*\n+ * For each 'pick' command, find out if the commit has a decoration in\n+ * refs/heads/. If so, then add a 'label for-update-refs/' command.\n+ */\n+static int todo_list_add_update_ref_commands(struct todo_list *todo_list)\n+{\n+\tint i;\n+\tstatic struct string_list decorate_refs_exclude = STRING_LIST_INIT_NODUP;\n+\tstatic struct string_list decorate_refs_exclude_config = STRING_LIST_INIT_NODUP;\n+\tstatic struct string_list decorate_refs_include = STRING_LIST_INIT_NODUP;\n+\tstruct decoration_filter decoration_filter = {\n+\t\t.include_ref_pattern = &decorate_refs_include,\n+\t\t.exclude_ref_pattern = &decorate_refs_exclude,\n+\t\t.exclude_ref_config_pattern = &decorate_refs_exclude_config\n+\t};\n+\tstruct todo_add_branch_context ctx = {\n+\t\t.buf = &todo_list->buf,\n+\t};\n+\n+\tctx.items_alloc = 2 * todo_list->nr + 1;\n+\tALLOC_ARRAY(ctx.items, ctx.items_alloc);\n+\n+\tstring_list_append(&decorate_refs_include, \"refs/heads/\");\n+\tload_ref_decorations(&decoration_filter, 0);\n+\n+\tfor (i = 0; i < todo_list->nr; ) {\n+\t\tstruct todo_item *item = &todo_list->items[i];\n+\n+\t\t/* insert ith item into new list */\n+\t\tALLOC_GROW(ctx.items,\n+\t\t\t ctx.items_nr + 1,\n+\t\t\t ctx.items_alloc);\n+\n+\t\tctx.items[ctx.items_nr++] = todo_list->items[i++];\n+\n+\t\tif (item->commit) {\n+\t\t\tctx.commit = item->commit;\n+\t\t\tfor_each_decoration(item->commit,\n+\t\t\t\t\t add_branch_for_decoration,\n+\t\t\t\t\t &ctx);\n+\t\t}\n+\t}\n+\n+\t/* Add the \"update-refs\" step. */\n+\tALLOC_GROW(ctx.items,\n+\t\t ctx.items_nr + 1,\n+\t\t ctx.items_alloc);\n+\tmemset(&ctx.items[ctx.items_nr], 0, sizeof(struct todo_item));\n+\tctx.items[ctx.items_nr].command = TODO_UPDATE_REFS;\n+\tctx.items_nr++;\n+\n+\tfree(todo_list->items);\n+\ttodo_list->items = ctx.items;\n+\ttodo_list->nr = ctx.items_nr;\n+\ttodo_list->alloc = ctx.items_alloc;\n+\n+\treturn 0;\n+}\n+\n int complete_action(struct repository *r, struct replay_opts *opts, unsigned flags,\n \t\t const char *shortrevisions, const char *onto_name,\n \t\t struct commit *onto, const struct object_id *orig_head,\n \t\t struct string_list *commands, unsigned autosquash,\n+\t\t unsigned update_refs,\n \t\t struct todo_list *todo_list)\n {\n \tchar shortonto[GIT_MAX_HEXSZ + 1];\n@@ -5636,6 +5738,9 @@ int complete_action(struct repository *r, struct replay_opts *opts, unsigned fla\n \t\titem->arg_len = item->arg_offset = item->flags = item->offset_in_buf = 0;\n \t}\n \n+\tif (update_refs && todo_list_add_update_ref_commands(todo_list))\n+\t\treturn -1;\n+\n \tif (autosquash && todo_list_rearrange_squash(todo_list))\n \t\treturn -1;\n \ndiff --git a/sequencer.h b/sequencer.h\nindex c2b4e148d8f..e268208b315 100644\n--- a/sequencer.h\n+++ b/sequencer.h\n@@ -167,6 +167,7 @@ int complete_action(struct repository *r, struct replay_opts *opts, unsigned fla\n \t\t const char *shortrevisions, const char *onto_name,\n \t\t struct commit *onto, const struct object_id *orig_head,\n \t\t struct string_list *commands, unsigned autosquash,\n+\t\t unsigned update_refs,\n \t\t struct todo_list *todo_list);\n int todo_list_rearrange_squash(struct todo_list *todo_list);\n \ndiff --git a/t/t3404-rebase-interactive.sh b/t/t3404-rebase-interactive.sh\nindex f31afd4a547..5e99ad7f3b6 100755\n--- a/t/t3404-rebase-interactive.sh\n+++ b/t/t3404-rebase-interactive.sh\n@@ -1743,6 +1743,78 @@ test_expect_success 'ORIG_HEAD is updated correctly' '\n \ttest_cmp_rev ORIG_HEAD test-orig-head@{1}\n '\n \n+test_expect_success '--update-refs adds label and update-ref commands' '\n+\tgit checkout -b update-refs no-conflict-branch &&\n+\tgit branch -f base HEAD~4 &&\n+\tgit branch -f first HEAD~3 &&\n+\tgit branch -f second HEAD~3 &&\n+\tgit branch -f third HEAD~1 &&\n+\tgit commit --allow-empty --fixup=third &&\n+\tgit branch -f shared-tip &&\n+\t(\n+\t\tset_cat_todo_editor &&\n+\n+\t\tcat >expect <<-EOF &&\n+\t\tpick $(git log -1 --format=%h J) J\n+\t\tlabel for-update-refs/refs/heads/second\n+\t\tlabel for-update-refs/refs/heads/first\n+\t\tpick $(git log -1 --format=%h K) K\n+\t\tpick $(git log -1 --format=%h L) L\n+\t\tfixup $(git log -1 --format=%h update-refs) fixup! L # empty\n+\t\tlabel for-update-refs/refs/heads/third\n+\t\tpick $(git log -1 --format=%h M) M\n+\t\tlabel for-update-refs/refs/heads/no-conflict-branch\n+\t\tlabel for-update-refs/refs/heads/shared-tip\n+\t\tupdate-refs\n+\t\tEOF\n+\n+\t\ttest_must_fail git rebase -i --autosquash --update-refs primary >todo &&\n+\t\ttest_cmp expect todo\n+\t)\n+'\n+\n+test_expect_success '--update-refs adds commands with --rebase-merges' '\n+\tgit checkout -b update-refs-with-merge no-conflict-branch &&\n+\tgit branch -f base HEAD~4 &&\n+\tgit branch -f first HEAD~3 &&\n+\tgit branch -f second HEAD~3 &&\n+\tgit branch -f third HEAD~1 &&\n+\tgit merge -m merge branch2 &&\n+\tgit branch -f merge-branch &&\n+\tgit commit --fixup=third --allow-empty &&\n+\t(\n+\t\tset_cat_todo_editor &&\n+\n+\t\tcat >expect <<-EOF &&\n+\t\tlabel onto\n+\t\treset onto\n+\t\tpick $(git log -1 --format=%h branch2~1) F\n+\t\tpick $(git log -1 --format=%h branch2) I\n+\t\tlabel for-update-refs/refs/heads/branch2\n+\t\tlabel merge\n+\t\treset onto\n+\t\tpick $(git log -1 --format=%h refs/heads/second) J\n+\t\tlabel for-update-refs/refs/heads/second\n+\t\tlabel for-update-refs/refs/heads/first\n+\t\tpick $(git log -1 --format=%h refs/heads/third~1) K\n+\t\tpick $(git log -1 --format=%h refs/heads/third) L\n+\t\tfixup $(git log -1 --format=%h update-refs-with-merge) fixup! L # empty\n+\t\tlabel for-update-refs/refs/heads/third\n+\t\tpick $(git log -1 --format=%h HEAD~2) M\n+\t\tlabel for-update-refs/refs/heads/no-conflict-branch\n+\t\tmerge -C $(git log -1 --format=%h HEAD~1) merge # merge\n+\t\tlabel for-update-refs/refs/heads/merge-branch\n+\t\tupdate-refs\n+\t\tEOF\n+\n+\t\ttest_must_fail git rebase -i --autosquash \\\n+\t\t\t\t --rebase-merges=rebase-cousins \\\n+\t\t\t\t --update-refs primary >todo &&\n+\n+\t\ttest_cmp expect todo\n+\t)\n+'\n+\n # This must be the last test in this file\n test_expect_success '$EDITOR and friends are unchanged' '\n \ttest_editor_unchanged\n-- \ngitgitgadget\n\n"},{"id":"456850","messageId":"3d7d3f656b4e93e8caa0d18d29c318ede956d1d7.1654634569.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","subject":"[PATCH v2 7/7] rebase: add rebase.updateRefs config option","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-07T20:42:49Z","receivedAt":"2022-06-08T02:55:54Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe previous change added the --update-refs command-line option. For\nusers who always want this mode, create the rebase.updateRefs config\noption which behaves the same way as rebase.autoSquash does with the\n--autosquash option.\n\nSigned-off-by: Derrick Stolee \n---\n Documentation/config/rebase.txt | 3 +++\n Documentation/git-rebase.txt | 4 ++++\n builtin/rebase.c | 5 +++++\n t/t3404-rebase-interactive.sh | 14 ++++++++++++++\n 4 files changed, 26 insertions(+)\n\ndiff --git a/Documentation/config/rebase.txt b/Documentation/config/rebase.txt\nindex 8c979cb20f2..f19bd0e0407 100644\n--- a/Documentation/config/rebase.txt\n+++ b/Documentation/config/rebase.txt\n@@ -21,6 +21,9 @@ rebase.autoStash::\n \t`--autostash` options of linkgit:git-rebase[1].\n \tDefaults to false.\n \n+rebase.updateRefs::\n+\tIf set to true enable `--update-refs` option by default.\n+\n rebase.missingCommitsCheck::\n \tIf set to \"warn\", git rebase -i will print a warning if some\n \tcommits are removed (e.g. a line was deleted), however the\ndiff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt\nindex e7611b4089c..7a56dbb9bec 100644\n--- a/Documentation/git-rebase.txt\n+++ b/Documentation/git-rebase.txt\n@@ -615,6 +615,10 @@ start would be overridden by the presence of\n \tare being rebased. Any branches that are checked out in a worktree\n \tor point to a `squash! ...` or `fixup! ...` commit are not updated\n \tin this way.\n++\n+If the `--update-refs` option is enabled by default using the\n+configuration variable `rebase.updateRefs`, this option can be\n+used to override and disable this setting.\n \n INCOMPATIBLE OPTIONS\n --------------------\ndiff --git a/builtin/rebase.c b/builtin/rebase.c\nindex 56d82a52106..8ebc98ea505 100644\n--- a/builtin/rebase.c\n+++ b/builtin/rebase.c\n@@ -802,6 +802,11 @@ static int rebase_config(const char *var, const char *value, void *data)\n \t\treturn 0;\n \t}\n \n+\tif (!strcmp(var, \"rebase.updaterefs\")) {\n+\t\topts->update_refs = git_config_bool(var, value);\n+\t\treturn 0;\n+\t}\n+\n \tif (!strcmp(var, \"rebase.reschedulefailedexec\")) {\n \t\topts->reschedule_failed_exec = git_config_bool(var, value);\n \t\treturn 0;\ndiff --git a/t/t3404-rebase-interactive.sh b/t/t3404-rebase-interactive.sh\nindex 72711efec28..f580afd8723 100755\n--- a/t/t3404-rebase-interactive.sh\n+++ b/t/t3404-rebase-interactive.sh\n@@ -1769,6 +1769,12 @@ test_expect_success '--update-refs adds label and update-ref commands' '\n \t\tEOF\n \n \t\ttest_must_fail git rebase -i --autosquash --update-refs primary >todo &&\n+\t\ttest_cmp expect todo &&\n+\n+\t\ttest_must_fail git -c rebase.autosquash=true \\\n+\t\t\t\t -c rebase.updaterefs=true \\\n+\t\t\t\t rebase -i primary >todo &&\n+\n \t\ttest_cmp expect todo\n \t)\n '\n@@ -1811,6 +1817,14 @@ test_expect_success '--update-refs adds commands with --rebase-merges' '\n \t\t\t\t --rebase-merges=rebase-cousins \\\n \t\t\t\t --update-refs primary >todo &&\n \n+\t\ttest_cmp expect todo &&\n+\n+\t\ttest_must_fail git -c rebase.autosquash=true \\\n+\t\t\t\t -c rebase.updaterefs=true \\\n+\t\t\t\t rebase -i \\\n+\t\t\t\t --rebase-merges=rebase-cousins \\\n+\t\t\t\t primary >todo &&\n+\n \t\ttest_cmp expect todo\n \t)\n '\n-- \ngitgitgadget\n"},{"id":"456856","messageId":"e84b8f19-4fb4-77cd-7e56-087e84e59817@github.com","threadId":"57949","inReplyTo":"xmqqr140t9am.fsf@gitster.g","subject":"Re: [PATCH v2 2/7] branch: add branch_checked_out() helper","fromName":"Derrick Stolee","fromEmail":"derrickstolee@github.com","sentAt":"2022-06-08T02:14:16Z","receivedAt":"2022-06-08T05:17:19Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"On 6/7/2022 6:09 PM, Junio C Hamano wrote:\n> \"Derrick Stolee via GitGitGadget\" writes:\n> \n>> +int branch_checked_out(const char *refname, char **path)\n>> +{\n>> +\tstruct worktree **worktrees = get_worktrees();\n>> +\tconst struct worktree *wt = find_shared_symref(worktrees, \"HEAD\", refname);\n>> +\tint result = wt && !wt->is_bare;\n>> +\n>> +\tif (result && path)\n>> +\t\t*path = xstrdup(wt->path);\n>> +\n>> +\tfree_worktrees(worktrees);\n>> +\treturn result;\n>> +}\n> \n> Don't you plan to call this repeatedly from the for_each_deco\n> iteration? I am wondering if it should take the result of\n> get_worktrees() and reuse the result of get_worktrees(), instead of\n> enumerating the list of worktrees and freeing for each of the\n> branches you need to inspect.\n\nSorry, you had mentioned this in v1 and I just forgot to fix this.\n\n> There also was another topic\n> \n> https://lore.kernel.org/git/pull.1266.v2.git.git.1652690501963.gitgitgadget@gmail.com/\n> \n> that was triggered by find_shared_symref() being relatively\n> heavy-weight, which suggests a more involved refactoring.\n> \n> I wonder if we rather want to rewrite find_shared_symref() *not* to\n> take the target parameter at all, and instead introduce a new\n> function that takes a worktree, and report the branch that is\n> checked out (or being operated on via rebase or bisect). Then we\n> can\n> \n> - create a strset out of its result, i.e. set of branches that\n> should not be touched;\n> \n> - iterate over refs that point into the history being rebased\n> (using for_each_decoration()), and consult that strset to see if\n> any of them is being rewritten.\n> \n> With the API of find_shared_symref(), we'd need to iterate over all\n> worktrees for each decoration. With such a restructuring, we can\n> iterate over all worktrees just once, and match the result with\n> decoration, so the problem becomes O(N)+O(M) and not O(N*M) for\n> number of worktrees N and number of decorations M.\n\nYes, that's a good plan. I'll take a look.\n\nThanks,\n-Stolee\n"},{"id":"456857","messageId":"74d5f74b-c95f-6892-6a66-3e646b1f9e53@github.com","threadId":"57949","inReplyTo":"e84b8f19-4fb4-77cd-7e56-087e84e59817@github.com","subject":"Re: [PATCH v2 2/7] branch: add branch_checked_out() helper","fromName":"Derrick Stolee","fromEmail":"derrickstolee@github.com","sentAt":"2022-06-08T02:43:00Z","receivedAt":"2022-06-08T05:29:58Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"On 6/7/2022 10:14 PM, Derrick Stolee wrote:\n> On 6/7/2022 6:09 PM, Junio C Hamano wrote:\n>>\n>> I wonder if we rather want to rewrite find_shared_symref() *not* to\n>> take the target parameter at all, and instead introduce a new\n>> function that takes a worktree, and report the branch that is\n>> checked out (or being operated on via rebase or bisect). Then we\n>> can\n>>\n>> - create a strset out of its result, i.e. set of branches that\n>> should not be touched;\n>>\n>> - iterate over refs that point into the history being rebased\n>> (using for_each_decoration()), and consult that strset to see if\n>> any of them is being rewritten.\n>>\n>> With the API of find_shared_symref(), we'd need to iterate over all\n>> worktrees for each decoration. With such a restructuring, we can\n>> iterate over all worktrees just once, and match the result with\n>> decoration, so the problem becomes O(N)+O(M) and not O(N*M) for\n>> number of worktrees N and number of decorations M.\n> \n> Yes, that's a good plan. I'll take a look.\n\nHere is a fixup for this patch that should work. I'll put it in v3.\n\ndiff --git a/branch.c b/branch.c\nindex 2e6419cdfa5..514212f5619 100644\n--- a/branch.c\n+++ b/branch.c\n@@ -10,6 +10,7 @@\n #include \"worktree.h\"\n #include \"submodule-config.h\"\n #include \"run-command.h\"\n+#include \"strmap.h\"\n \n struct tracking {\n \tstruct refspec_item spec;\n@@ -369,17 +370,44 @@ int validate_branchname(const char *name, struct strbuf *ref)\n \treturn ref_exists(ref->buf);\n }\n \n-int branch_checked_out(const char *refname, char **path)\n+static int initialized_checked_out_branches;\n+static struct strmap current_checked_out_branches = STRMAP_INIT;\n+\n+static void prepare_checked_out_branches(void)\n {\n-\tstruct worktree **worktrees = get_worktrees();\n-\tconst struct worktree *wt = find_shared_symref(worktrees, \"HEAD\", refname);\n-\tint result = wt && !wt->is_bare;\n+\tint i = 0;\n+\tstruct worktree **worktrees;\n+\n+\tif (initialized_checked_out_branches)\n+\t\treturn;\n+\tinitialized_checked_out_branches = 1;\n+\n+\tworktrees = get_worktrees();\n+\n+\twhile (worktrees[i]) {\n+\t\tstruct worktree *wt = worktrees[i];\n \n-\tif (result && path)\n-\t\t*path = xstrdup(wt->path);\n+\t\tif (!wt->is_bare && wt->head_ref)\n+\t\t\tstrmap_put(¤t_checked_out_branches,\n+\t\t\t\t wt->head_ref,\n+\t\t\t\t wt->path);\n+\n+\t\ti++;\n+\t}\n \n \tfree_worktrees(worktrees);\n-\treturn result;\n+}\n+\n+int branch_checked_out(const char *refname, char **path)\n+{\n+\tconst char *path_in_set;\n+\tprepare_checked_out_branches();\n+\n+\tpath_in_set = strmap_get(¤t_checked_out_branches, refname);\n+\tif (path_in_set && path)\n+\t\t*path = xstrdup(path_in_set);\n+\n+\treturn !!path_in_set;\n }\n \n /*\n\n>> There also was another topic\n>>\n>> https://lore.kernel.org/git/pull.1266.v2.git.git.1652690501963.gitgitgadget@gmail.com/\n>>\n>> that was triggered by find_shared_symref() being relatively\n>> heavy-weight, which suggests a more involved refactoring.\n\nThe patch there was this:\n\ndiff --git a/builtin/fetch.c b/builtin/fetch.c\nindex e3791f09ed5..eeee5ac8f15 100644\n--- a/builtin/fetch.c\n+++ b/builtin/fetch.c\n@@ -1440,6 +1440,7 @@ static void check_not_current_branch(struct ref *ref_map,\n \tconst struct worktree *wt;\n \tfor (; ref_map; ref_map = ref_map->next)\n \t\tif (ref_map->peer_ref &&\n+\t\t starts_with(ref_map->peer_ref->name, \"refs/heads/\") &&\n \t\t (wt = find_shared_symref(worktrees, \"HEAD\",\n \t\t\t\t\t ref_map->peer_ref->name)) &&\n \t\t !wt->is_bare)\n\nAnd there is another use of find_shared_symref() in the same file, allowing\nus to do the following:\n\ndiff --git a/builtin/fetch.c b/builtin/fetch.c\nindex ac29c2b1ae3..3933c482839 100644\n--- a/builtin/fetch.c\n+++ b/builtin/fetch.c\n@@ -885,7 +885,7 @@ static int update_local_ref(struct ref *ref,\n \t\t\t struct worktree **worktrees)\n {\n \tstruct commit *current = NULL, *updated;\n-\tconst struct worktree *wt;\n+\tchar *path = NULL;\n \tconst char *pretty_ref = prettify_refname(ref->name);\n \tint fast_forward = 0;\n \n@@ -900,17 +900,17 @@ static int update_local_ref(struct ref *ref,\n \t}\n \n \tif (!update_head_ok &&\n-\t (wt = find_shared_symref(worktrees, \"HEAD\", ref->name)) &&\n-\t !wt->is_bare && !is_null_oid(&ref->old_oid)) {\n+\t !is_null_oid(&ref->old_oid) &&\n+\t branch_checked_out(ref->name, &path)) {\n \t\t/*\n \t\t * If this is the head, and it's not okay to update\n \t\t * the head, and the old value of the head isn't empty...\n \t\t */\n \t\tformat_display(display, '!', _(\"[rejected]\"),\n-\t\t\t wt->is_current ?\n-\t\t\t\t _(\"can't fetch in current branch\") :\n-\t\t\t\t _(\"checked out in another worktree\"),\n+\t\t\t path ? _(\"can't fetch in current branch\") :\n+\t\t\t\t _(\"checked out in another worktree\"),\n \t\t\t remote, pretty_ref, summary_width);\n+\t\tfree(path);\n \t\treturn 1;\n \t}\n \n@@ -1437,16 +1437,14 @@ static int prune_refs(struct refspec *rs,\n static void check_not_current_branch(struct ref *ref_map,\n \t\t\t\t struct worktree **worktrees)\n {\n-\tconst struct worktree *wt;\n+\tchar *path;\n \tfor (; ref_map; ref_map = ref_map->next)\n \t\tif (ref_map->peer_ref &&\n \t\t starts_with(ref_map->peer_ref->name, \"refs/heads/\") &&\n-\t\t (wt = find_shared_symref(worktrees, \"HEAD\",\n-\t\t\t\t\t ref_map->peer_ref->name)) &&\n-\t\t !wt->is_bare)\n+\t\t branch_checked_out(ref_map->peer_ref->name, &path))\n \t\t\tdie(_(\"refusing to fetch into branch '%s' \"\n \t\t\t \"checked out at '%s'\"),\n-\t\t\t ref_map->peer_ref->name, wt->path);\n+\t\t\t ref_map->peer_ref->name, path);\n }\n \n static int truncate_fetch_head(void)\n\nI can extract these as patches in their own series if we think\nthat's something we worth fast-tracking.\n\nThanks,\n-Stolee\n"},{"id":"456858","messageId":"xmqqleu7sqnt.fsf@gitster.g","threadId":"57949","inReplyTo":"74d5f74b-c95f-6892-6a66-3e646b1f9e53@github.com","subject":"Re: [PATCH v2 2/7] branch: add branch_checked_out() helper","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-08T04:52:22Z","receivedAt":"2022-06-08T06:55:46Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Derrick Stolee writes:\n\n>> Yes, that's a good plan. I'll take a look.\n>\n> Here is a fixup for this patch that should work. I'll put it in v3.\n>\n> diff --git a/branch.c b/branch.c\n> index 2e6419cdfa5..514212f5619 100644\n> --- a/branch.c\n> +++ b/branch.c\n> @@ -10,6 +10,7 @@\n> #include \"worktree.h\"\n> #include \"submodule-config.h\"\n> #include \"run-command.h\"\n> +#include \"strmap.h\"\n> \n> struct tracking {\n> \tstruct refspec_item spec;\n> @@ -369,17 +370,44 @@ int validate_branchname(const char *name, struct strbuf *ref)\n> \treturn ref_exists(ref->buf);\n> }\n> \n> -int branch_checked_out(const char *refname, char **path)\n> +static int initialized_checked_out_branches;\n> +static struct strmap current_checked_out_branches = STRMAP_INIT;\n> +\n> +static void prepare_checked_out_branches(void)\n> {\n> +\tint i = 0;\n> +\tstruct worktree **worktrees;\n> +\n> +\tif (initialized_checked_out_branches)\n> +\t\treturn;\n> +\tinitialized_checked_out_branches = 1;\n> +\n> +\tworktrees = get_worktrees();\n> +\n> +\twhile (worktrees[i]) {\n> +\t\tstruct worktree *wt = worktrees[i];\n> \n> +\t\tif (!wt->is_bare && wt->head_ref)\n> +\t\t\tstrmap_put(¤t_checked_out_branches,\n> +\t\t\t\t wt->head_ref,\n> +\t\t\t\t wt->path);\n> +\n> +\t\ti++;\n> +\t}\n> \tfree_worktrees(worktrees);\n> -\treturn result;\n> +}\n\nYeah, the above illustrates what I had in mind pretty closely. The\nabove outline only checks where the HEAD symref points at, and it\ndoes not protect a branch that is in the middle of getting bisected\nor rebased, which find_shared_symref() tries to do, IIRC, though.\n\nIt should not be too hard to add that to the above to allow us to\nachieve feature parity with the original.\n\n> +\n> +int branch_checked_out(const char *refname, char **path)\n> +{\n> +\tconst char *path_in_set;\n> +\tprepare_checked_out_branches();\n> +\n> +\tpath_in_set = strmap_get(¤t_checked_out_branches, refname);\n> +\tif (path_in_set && path)\n> +\t\t*path = xstrdup(path_in_set);\n> +\n> +\treturn !!path_in_set;\n> }\n\nThis one looks quite straight-forward.\n\n> And there is another use of find_shared_symref() in the same file, allowing\n> us to do the following:\n>\n> diff --git a/builtin/fetch.c b/builtin/fetch.c\n> index ac29c2b1ae3..3933c482839 100644\n> --- a/builtin/fetch.c\n> +++ b/builtin/fetch.c\n> @@ -885,7 +885,7 @@ static int update_local_ref(struct ref *ref,\n> \t\t\t struct worktree **worktrees)\n> {\n> \tstruct commit *current = NULL, *updated;\n> -\tconst struct worktree *wt;\n> +\tchar *path = NULL;\n> \tconst char *pretty_ref = prettify_refname(ref->name);\n> \tint fast_forward = 0;\n> \n> @@ -900,17 +900,17 @@ static int update_local_ref(struct ref *ref,\n> \t}\n> \n> \tif (!update_head_ok &&\n> -\t (wt = find_shared_symref(worktrees, \"HEAD\", ref->name)) &&\n> -\t !wt->is_bare && !is_null_oid(&ref->old_oid)) {\n> +\t !is_null_oid(&ref->old_oid) &&\n> +\t branch_checked_out(ref->name, &path)) {\n\nYes. It looks like a good direction to go in.\n\nThanks.\n"},{"id":"456863","messageId":"ac135ece-a901-5868-f28c-2161d7a62e97@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","subject":"Re: [PATCH v2 0/7] rebase: update branches in multi-part topic","fromName":"Phillip Wood","fromEmail":"phillip.wood123@gmail.com","sentAt":"2022-06-08T14:32:46Z","receivedAt":"2022-06-08T14:32:56Z","isPatch":true,"sender":{"key":"phillip.wood@dunelm.org.uk","avatar":null},"body":"Hi Stolee\n\nOn 07/06/2022 21:42, Derrick Stolee via GitGitGadget wrote:\n> [...]\n> As an example, here is my in-progress bundle URI RFC split into subtopics as\n> they appear during the TODO list of a git rebase -i --update-refs:\n> \n> pick 2d966282ff3 docs: document bundle URI standard\n> pick 31396e9171a remote-curl: add 'get' capability\n> pick 54c6ab70f67 bundle-uri: create basic file-copy logic\n> pick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\n> pick 6adaf842684 fetch: add --bundle-uri option\n> pick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\n> label for-update-refs/refs/heads/bundle-redo/fetch\n> [...] \n> update-refs\n> [...]\n> Based on some of the discussion, it seemed like one way to do this would be\n> to have an 'update-ref ' command that would take the place of these 'label'\n> commands. However, this would require two things that make it a bit awkward:\n> \n> 1. We would need to replicate the storage of those positions during the\n> rebase. 'label' already does this pretty well. I've added the\n> \"for-update-refs/\" label to help here.\n\nI'm afraid I don't think that using a label with a name constructed from \na magic prefix and the full refname is a good user interface.\n\n(i) Having labels behave differently based on their name is confusing.\n\n(ii) The length of the label string is cumbersome for users. Rebase \nalready has a reputation for being unfriendly and hard to use, this will \nnot improve that. \"update-ref bundle-redo/fetch\" is much simpler.\n\n(iii) It means that we no longer store the oid of each branch when \ncreating the todo list and so cannot check if it is safe to update it at \nthe end of the rebase (just using the current value of the branch ref at \nthe end of a long running operation like rebase is not sufficient to be \nsafe).\n\n> 2. If we want to close out all of the refs as the rebase is finishing, then\n> that \"step\" becomes invisible to the user (and a bit more complicated to\n> insert). Thus, the 'update-refs' step performs this action. If the user\n> wants to do things after that step, then they can do so by editing the\n> TODO list.\n\nI'm not sure what the use case is for making the update-refs step \nvisible to the user - it seems to be added complexity for them to deal \nwith. We don't do that for the branch that is being rebased so why do we \nneed to do it for the other branches? As for the implementation we could \njust update the branches at the end of the loop in pick_commits() where \nwe update the branch and run the post-rewrite hook etc. there is no need \nfor an entry in the todo list.\n\nBest Wishes\n\nPhillip\n"},{"id":"456869","messageId":"xmqqbkv3rvlj.fsf@gitster.g","threadId":"57949","inReplyTo":"9354d1d3-c1b7-3baf-215f-30659ad48b22@github.com","subject":"Re: [PATCH 0/4] rebase: update branches in multi-part topic","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2022-06-08T16:03:20Z","receivedAt":"2022-06-08T16:03:47Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Derrick Stolee writes:\n\n>> I think the question of whether to update branches that are checked out in\n>> another worktree is a question of whether it is less inconvenient to the user\n>> to skip the ref update and leave the user to manually update the branch or to\n>> update the ref and leave the worktree in a potentially awkward state if the\n>> user was half way through building a commit. The answer probably depends on\n>> the preferences of the user.\n\nTo some degree, it might be true, but I think the recent thinking is\nthat by default any refs being worked on by the user must be kept\nintact. \"switch\" does not let you check out a branch that is\nalready checked out elsewhere, \"fetch\" does not let you overwrite\nthe branch that is currently checked out without \"--update-head-ok\",\netc.\n\n> I think that their 'git status' will look strange no matter what: their\n> working directory and index could look significantly different from what\n> the branch at HEAD is reporting. For this situation, I would rather continue\n> preventing these ref updates from underneath worktrees.\n> \n>> I've been using a script that updates the refs for all the branches being\n>> rewritten for a while and have found it preferable to always update the ref\n>> rather than have to do it manually. My script also updates the worktree\n>> checkout to the new HEAD if there are no uncommitted changes which I have\n>> found very convenient. My preference is probably because I tend not to have\n>> uncommitted changes lying around in the worktrees whose branches get updated.\n>\n> Actually updating the worktree to match seems like an interesting twist, which\n> we would want to consider if we go this route in the future.\n\nUsually I caution against adding \"features\" that can complicate the\nend-user experience like this, but in this case, the potential for\nextra conflicts coming from such updates to the working trees may\nnot be too bad.\n\nIt all depends on why the user has these intermediate branches and\nhow they are used, but in order to see any conflicts happen after\nrebasing a branch here, before you start that rebase, you need to\nhave in different working trees that had these intermediate branches\nchecked out and had local modifications on top of them.\n\nI would say that you deserve it if conflicts resulting from such a\nset-up hurts you ;-) If you are using these intermediate branches\nso that you can work on steps in the middle of the larger whole,\nthen the proposed \"rebase --update-refs\" is not for you. Having\nlocal modification in these separate working trees that check out\nthe intermediate branches is OK, but once you create a commit on\nsuch a branch, the commit that is in the larger topic is no longer\nat the tip, so \"rebase --update-refs\" would not notice it anyway.\n\nRather, if you have the intermediate branches, you'd probably want\nto work on each of them to make them stable, and rebase them on top\nof each other in an appropriate order. \"rebase --update-refs\" that\nruns on the larger topic does not have enough information to rebuild\nit with tips of intermediate branches that are updated elsewhere.\n\nWhich means that anybody sane who uses \"rebase --update-refs\" would\nnot modify these intermediate branches outside the context of the\nlarger topic, and those who do not follow this rule can keep both\nhalves of their history ;-)\n\nFWIW a complementary tool that would work in the other direction, to\nexpect the user to have worked on smaller branches and rebuild the\nlarger branch that contains them, is also missing from our tool set.\nThose who check out these smaller branches in separate working trees\nwould want to use such a tool, not \"rebase --update-ref\".\n\nThanks.\n"},{"id":"456879","messageId":"34264915-8a37-62db-f954-0b5297639b34@github.com","threadId":"57949","inReplyTo":"ac135ece-a901-5868-f28c-2161d7a62e97@gmail.com","subject":"Re: [PATCH v2 0/7] rebase: update branches in multi-part topic","fromName":"Derrick Stolee","fromEmail":"derrickstolee@github.com","sentAt":"2022-06-08T18:09:01Z","receivedAt":"2022-06-08T18:09:11Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"On 6/8/2022 10:32 AM, Phillip Wood wrote:\n> Hi Stolee\n> \n> On 07/06/2022 21:42, Derrick Stolee via GitGitGadget wrote:\n>> [...]\n>> As an example, here is my in-progress bundle URI RFC split into subtopics as\n>> they appear during the TODO list of a git rebase -i --update-refs:\n>>\n>> pick 2d966282ff3 docs: document bundle URI standard\n>> pick 31396e9171a remote-curl: add 'get' capability\n>> pick 54c6ab70f67 bundle-uri: create basic file-copy logic\n>> pick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\n>> pick 6adaf842684 fetch: add --bundle-uri option\n>> pick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\n>> label for-update-refs/refs/heads/bundle-redo/fetch\n>> [...] update-refs\n>> [...]\n>> Based on some of the discussion, it seemed like one way to do this would be\n>> to have an 'update-ref ' command that would take the place of these 'label'\n>> commands. However, this would require two things that make it a bit awkward:\n>>\n>>   1. We would need to replicate the storage of those positions during the\n>>      rebase. 'label' already does this pretty well. I've added the\n>>      \"for-update-refs/\" label to help here.\n> \n> I'm afraid I don't think that using a label with a name constructed from\n> a magic prefix and the full refname is a good user interface.\n> \n> (i) Having labels behave differently based on their name is confusing.\n\nThe label commands do as advertised, but the 'update-refs' command does\nsomething with the set of refs based on their names, yes. We would need to\nstore this information during the rebase _somewhere_, and refs/rewritten/\nseems natural.\n \n> (ii) The length of the label string is cumbersome for users. Rebase already\n> has a reputation for being unfriendly and hard to use, this will not improve\n> that. \"update-ref bundle-redo/fetch\" is much simpler.\n\nI agree that your proposed replacement for \"label for-update-refs/...\" would\nlook cleaner. I don't think that that is enough to justify hiding information\nfrom the user.\n\n> (iii) It means that we no longer store the oid of each branch when creating> the todo list and so cannot check if it is safe to update it at the end of the\n> rebase (just using the current value of the branch ref at the end of a long\n> running operation like rebase is not sufficient to be safe).\n\nThe operation we are doing necessarily requires a force-update, since we are\nmodifying history. The safety you are caring about here is about the case where\nthe user modifies one of these refs between the initial 'git rebase --update-refs'\nand the final 'git rebase --continue' that actually writes the refs. You want to\nprevent that final update from succeeding because another change to those refs\ncould be lost.\n\nI'm less concerned about this case, because the user is requesting that we\nupdate the refs pointing to this set of commits. But, I'm glad you brought it\nup.\n\nOne way to prevent this situation would be to extend the idea of \"what branch\nis being rebased?\" (REBASE_HEAD) to \"which branches are being rewritten?\"\n(REBASE_HEADS?). If the worktree kept the full list somewhere in the $GIT_DIR,\nthen other Git commands could notice that those refs are currently being\noverwritten and those updates should fail (similarly to how we currently fail\nto update a branch being rebased). This ties into your later point:\n\n>>   2. If we want to close out all of the refs as the rebase is finishing, then\n>>      that \"step\" becomes invisible to the user (and a bit more complicated to\n>>      insert). Thus, the 'update-refs' step performs this action. If the user\n>>      wants to do things after that step, then they can do so by editing the\n>>      TODO list.\n> \n> I'm not sure what the use case is for making the update-refs step visible to\n> the user - it seems to be added complexity for them to deal with. We don't do\n> that for the branch that is being rebased so why do we need to do it for the\n> other branches? As for the implementation we could just update the branches\n> at the end of the loop in pick_commits() where we update the branch and run the\n> post-rewrite hook etc. there is no need for an entry in the todo list.\n\nI concede that allowing the user to move the 'update-refs' command around is\nnot super important.\n\nThe biggest concern I originally had with this idea was that there was no\nconnection between the \"--update-refs\" option in the first \"git rebase\"\ncommand and the final \"git rebase --continue\" command. That is, other than\nwhich refs are in refs/rewritten/*.\n\nHOWEVER, using refs/rewritten/* is likely buggy: if we had two rebases\nhappening at the same time (operating on a disjoint set of refs), then how\ndo we differentiate which refs make sense for us to update as we complete\nthis rebase?\n\nSo, I'm coming to the conclusion that using refs/rewritten/* is problematic\nand I should use something closer to REBASE_HEAD as a way to describe which\nrefs are being updated. In that context, your 'update-ref ' command\nmakes a lot more sense. The user can decide to move those around _or_ remove\nthem; it won't change their protection under \"REBASE_HEADS\" but would\nprevent them from being rewritten.\n\nWhile I think on this, I'll send my branch_checked_out() patches in a\nseparate thread, since those have grown in complexity since this version.\n\nThanks,\n-Stolee\n"},{"id":"456930","messageId":"2e6affba-f9ba-56da-20d6-009c5f2fbfd7@gmail.com","threadId":"57949","inReplyTo":"34264915-8a37-62db-f954-0b5297639b34@github.com","subject":"Re: [PATCH v2 0/7] rebase: update branches in multi-part topic","fromName":"Phillip Wood","fromEmail":"phillip.wood123@gmail.com","sentAt":"2022-06-09T10:04:07Z","receivedAt":"2022-06-09T10:04:16Z","isPatch":true,"sender":{"key":"phillip.wood@dunelm.org.uk","avatar":null},"body":"Hi Stolee\n\nOn 08/06/2022 19:09, Derrick Stolee wrote:\n> On 6/8/2022 10:32 AM, Phillip Wood wrote:\n>> Hi Stolee\n>>\n>> On 07/06/2022 21:42, Derrick Stolee via GitGitGadget wrote:\n>>> [...]\n>>> As an example, here is my in-progress bundle URI RFC split into subtopics as\n>>> they appear during the TODO list of a git rebase -i --update-refs:\n>>>\n>>> pick 2d966282ff3 docs: document bundle URI standard\n>>> pick 31396e9171a remote-curl: add 'get' capability\n>>> pick 54c6ab70f67 bundle-uri: create basic file-copy logic\n>>> pick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\n>>> pick 6adaf842684 fetch: add --bundle-uri option\n>>> pick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\n>>> label for-update-refs/refs/heads/bundle-redo/fetch\n>>> [...] update-refs\n>>> [...]\n>>> Based on some of the discussion, it seemed like one way to do this would be\n>>> to have an 'update-ref ' command that would take the place of these 'label'\n>>> commands. However, this would require two things that make it a bit awkward:\n>>>\n>>>   1. We would need to replicate the storage of those positions during the\n>>>      rebase. 'label' already does this pretty well. I've added the\n>>>      \"for-update-refs/\" label to help here.\n>>\n>> I'm afraid I don't think that using a label with a name constructed from\n>> a magic prefix and the full refname is a good user interface.\n>>\n>> (i) Having labels behave differently based on their name is confusing.\n> \n> The label commands do as advertised, but the 'update-refs' command does\n> something with the set of refs based on their names, yes. We would need to\n> store this information during the rebase _somewhere_, and refs/rewritten/\n> seems natural.\n> \n>> (ii) The length of the label string is cumbersome for users. Rebase already\n>> has a reputation for being unfriendly and hard to use, this will not improve\n>> that. \"update-ref bundle-redo/fetch\" is much simpler.\n> \n> I agree that your proposed replacement for \"label for-update-refs/...\" would\n> look cleaner. I don't think that that is enough to justify hiding information\n> from the user.\n\nIn general I think it is better not to burden the user with unnecessary \ninformation - it makes the ui more complicated and exposes \nimplementation details which makes it harder to change the implementation.\n\n>> (iii) It means that we no longer store the oid of each branch when creating> the todo list and so cannot check if it is safe to update it at the end of the\n>> rebase (just using the current value of the branch ref at the end of a long\n>> running operation like rebase is not sufficient to be safe).\n> \n> The operation we are doing necessarily requires a force-update, since we are\n> modifying history. The safety you are caring about here is about the case where\n> the user modifies one of these refs between the initial 'git rebase --update-refs'\n> and the final 'git rebase --continue' that actually writes the refs. You want to\n> prevent that final update from succeeding because another change to those refs\n> could be lost.\n> \n> I'm less concerned about this case, because the user is requesting that we\n> update the refs pointing to this set of commits. But, I'm glad you brought it\n> up.\n\nAt the end of the rebase we pass the oid of the branch that we store at \nthe start of the rebase when updating the ref to avoid nasty surprises. \nI think it makes sense to do the same for the other branches being \nrewritten - it is easy to get stuck on a conflict resolution and go and \ndo something else for a while and forget a branch is being rebased. If \nwe cannot update the ref at the end of the rebase we should print the \nnew oid with instructions to run \"git reset --hard\" or \"git update-ref\" \nif the user wants to update the branch.\n\n> One way to prevent this situation would be to extend the idea of \"what branch\n> is being rebased?\" (REBASE_HEAD) to \"which branches are being rewritten?\"\n> (REBASE_HEADS?). \n\nJust to clarify REBASE_HEAD points to the commit we're currently trying \nto pick, the branch ref is stored it .git/rebase-merge/head-name and its \noid in .git/rebase-merge/orig-head\n\n> If the worktree kept the full list somewhere in the $GIT_DIR,\n> then other Git commands could notice that those refs are currently being\n> overwritten and those updates should fail (similarly to how we currently fail\n> to update a branch being rebased).\n\nThat's a good point and one that I should have remembered as I \nimplemented that check in my script. I agree that we should stop the \nuser starting a rebase in a worktree whose branch is being updated \nelsewhere. If we write a list of the refs we want to update under \n.git/rebase-merge before walking the other worktrees to see what's being \nrebased elsewhere that avoids a race where two processes start that try \nto rebase the same branch in different worktrees at the same time. It \nwould be easy to store the original oids with the ref names.\n\nAn added complexity is that we would have to check the todo list for any \nnew update-ref commands and check those refs are not being rebased \nelsewhere each time the list is edited. In the future we should update \n\"git status\" to read that file but I don't think that needs to be part \nof the initial implementation.\n\n> This ties into your later point:\n> \n>>>   2. If we want to close out all of the refs as the rebase is finishing, then\n>>>      that \"step\" becomes invisible to the user (and a bit more complicated to\n>>>      insert). Thus, the 'update-refs' step performs this action. If the user\n>>>      wants to do things after that step, then they can do so by editing the\n>>>      TODO list.\n>>\n>> I'm not sure what the use case is for making the update-refs step visible to\n>> the user - it seems to be added complexity for them to deal with. We don't do\n>> that for the branch that is being rebased so why do we need to do it for the\n>> other branches? As for the implementation we could just update the branches\n>> at the end of the loop in pick_commits() where we update the branch and run the\n>> post-rewrite hook etc. there is no need for an entry in the todo list.\n> \n> I concede that allowing the user to move the 'update-refs' command around is\n> not super important.\n> \n> The biggest concern I originally had with this idea was that there was no\n> connection between the \"--update-refs\" option in the first \"git rebase\"\n> command and the final \"git rebase --continue\" command. That is, other than\n> which refs are in refs/rewritten/*.\n> \n> HOWEVER, using refs/rewritten/* is likely buggy: if we had two rebases\n> happening at the same time (operating on a disjoint set of refs), then how\n> do we differentiate which refs make sense for us to update as we complete\n> this rebase?\n >> So, I'm coming to the conclusion that using refs/rewritten/* is \nproblematic\n> and I should use something closer to REBASE_HEAD as a way to describe which\n> refs are being updated. In that context, your 'update-ref ' command\n> makes a lot more sense. The user can decide to move those around _or_ remove\n> them; it won't change their protection under \"REBASE_HEADS\" but would\n> prevent them from being rewritten.\n\nWhenever one adds a new option to rebase it almost always involves \nwriting more state to .git/rebase-merge, in this case I think we want to \nstore the branches from the update-ref commands and their original oids \nthere. I'm not sure that we want to expose the file to the user though \nso we can change the implementation in the future if we need to (e.g. we \nmay decide it is better to store this information for all worktrees \nunder a single file in $GIT_COMMON_DIR).\n\n> While I think on this, I'll send my branch_checked_out() patches in a\n> separate thread, since those have grown in complexity since this version.\n\nIf you want to discuss any ideas face-to-face drop me a email and we can \narrange a time to chat if that would be useful.\n\nBest Wishes\n\nPhillip\n\n> -Stolee\n"},{"id":"458035","messageId":"fbaedc7f1f02efec251cb4df665fd3cb71f1d3b3.1656422759.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v3.git.1656422759.gitgitgadget@gmail.com","subject":"[PATCH v3 1/8] t2407: test branches currently using apply backend","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-28T13:25:52Z","receivedAt":"2022-06-28T13:28:09Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe tests in t2407 that verify the branch_checked_out() helper in the\ncase of bisects and rebases were added by 9347303db89 (branch: check for\nbisects and rebases, 2022-06-08). However, that commit failed to check\nfor rebases that are using the 'apply' backend.\n\nAdd a test that checks the apply backend. The implementation was already\ncorrect here, but it is good to have regression tests before modifying\nthe implementation further.\n\nSigned-off-by: Derrick Stolee \n---\n t/t2407-worktree-heads.sh | 13 ++++++++++++-\n 1 file changed, 12 insertions(+), 1 deletion(-)\n\ndiff --git a/t/t2407-worktree-heads.sh b/t/t2407-worktree-heads.sh\nindex b6be42f74a2..4f59bc21303 100755\n--- a/t/t2407-worktree-heads.sh\n+++ b/t/t2407-worktree-heads.sh\n@@ -54,7 +54,18 @@ test_expect_success 'refuse to overwrite: worktree in bisect' '\n \tgrep \"cannot force update the branch '\\''fake-2'\\'' checked out at.*wt-4\" err\n '\n \n-test_expect_success 'refuse to overwrite: worktree in rebase' '\n+test_expect_success 'refuse to overwrite: worktree in rebase (apply)' '\n+\ttest_when_finished rm -rf .git/worktrees/wt-*/rebase-apply &&\n+\n+\tmkdir -p .git/worktrees/wt-3/rebase-apply &&\n+\techo refs/heads/fake-1 >.git/worktrees/wt-3/rebase-apply/head-name &&\n+\techo refs/heads/fake-2 >.git/worktrees/wt-3/rebase-apply/onto &&\n+\n+\ttest_must_fail git branch -f fake-1 HEAD 2>err &&\n+\tgrep \"cannot force update the branch '\\''fake-1'\\'' checked out at.*wt-3\" err\n+'\n+\n+test_expect_success 'refuse to overwrite: worktree in rebase (merge)' '\n \ttest_when_finished rm -rf .git/worktrees/wt-*/rebase-merge &&\n \n \tmkdir -p .git/worktrees/wt-3/rebase-merge &&\n-- \ngitgitgadget\n\n"},{"id":"458036","messageId":"d2cfdbfc4319081d548d13bbd675c4385b1a5ba6.1656422759.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v3.git.1656422759.gitgitgadget@gmail.com","subject":"[PATCH v3 8/8] rebase: add rebase.updateRefs config option","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-28T13:25:59Z","receivedAt":"2022-06-28T13:28:14Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe previous change added the --update-refs command-line option. For\nusers who always want this mode, create the rebase.updateRefs config\noption which behaves the same way as rebase.autoSquash does with the\n--autosquash option.\n\nSigned-off-by: Derrick Stolee \n---\n Documentation/config/rebase.txt | 3 +++\n Documentation/git-rebase.txt | 3 +++\n builtin/rebase.c | 5 +++++\n t/t3404-rebase-interactive.sh | 14 ++++++++++++++\n 4 files changed, 25 insertions(+)\n\ndiff --git a/Documentation/config/rebase.txt b/Documentation/config/rebase.txt\nindex 8c979cb20f2..f19bd0e0407 100644\n--- a/Documentation/config/rebase.txt\n+++ b/Documentation/config/rebase.txt\n@@ -21,6 +21,9 @@ rebase.autoStash::\n \t`--autostash` options of linkgit:git-rebase[1].\n \tDefaults to false.\n \n+rebase.updateRefs::\n+\tIf set to true enable `--update-refs` option by default.\n+\n rebase.missingCommitsCheck::\n \tIf set to \"warn\", git rebase -i will print a warning if some\n \tcommits are removed (e.g. a line was deleted), however the\ndiff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt\nindex e7611b4089c..1249f9ed617 100644\n--- a/Documentation/git-rebase.txt\n+++ b/Documentation/git-rebase.txt\n@@ -615,6 +615,9 @@ start would be overridden by the presence of\n \tare being rebased. Any branches that are checked out in a worktree\n \tor point to a `squash! ...` or `fixup! ...` commit are not updated\n \tin this way.\n++\n+If the configuration variable `rebase.updateRefs` is set, then this option\n+can be used to override and disable this setting.\n \n INCOMPATIBLE OPTIONS\n --------------------\ndiff --git a/builtin/rebase.c b/builtin/rebase.c\nindex 56d82a52106..8ebc98ea505 100644\n--- a/builtin/rebase.c\n+++ b/builtin/rebase.c\n@@ -802,6 +802,11 @@ static int rebase_config(const char *var, const char *value, void *data)\n \t\treturn 0;\n \t}\n \n+\tif (!strcmp(var, \"rebase.updaterefs\")) {\n+\t\topts->update_refs = git_config_bool(var, value);\n+\t\treturn 0;\n+\t}\n+\n \tif (!strcmp(var, \"rebase.reschedulefailedexec\")) {\n \t\topts->reschedule_failed_exec = git_config_bool(var, value);\n \t\treturn 0;\ndiff --git a/t/t3404-rebase-interactive.sh b/t/t3404-rebase-interactive.sh\nindex 63a69bc073e..77478d94f3b 100755\n--- a/t/t3404-rebase-interactive.sh\n+++ b/t/t3404-rebase-interactive.sh\n@@ -1768,6 +1768,12 @@ test_expect_success '--update-refs adds label and update-ref commands' '\n \t\tEOF\n \n \t\ttest_must_fail git rebase -i --autosquash --update-refs primary >todo &&\n+\t\ttest_cmp expect todo &&\n+\n+\t\ttest_must_fail git -c rebase.autosquash=true \\\n+\t\t\t\t -c rebase.updaterefs=true \\\n+\t\t\t\t rebase -i primary >todo &&\n+\n \t\ttest_cmp expect todo\n \t)\n '\n@@ -1809,6 +1815,14 @@ test_expect_success '--update-refs adds commands with --rebase-merges' '\n \t\t\t\t --rebase-merges=rebase-cousins \\\n \t\t\t\t --update-refs primary >todo &&\n \n+\t\ttest_cmp expect todo &&\n+\n+\t\ttest_must_fail git -c rebase.autosquash=true \\\n+\t\t\t\t -c rebase.updaterefs=true \\\n+\t\t\t\t rebase -i \\\n+\t\t\t\t --rebase-merges=rebase-cousins \\\n+\t\t\t\t primary >todo &&\n+\n \t\ttest_cmp expect todo\n \t)\n '\n-- \ngitgitgadget\n"},{"id":"458037","messageId":"918b398d6a2264a99acadd319f780c796bbafc18.1656422759.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v3.git.1656422759.gitgitgadget@gmail.com","subject":"[PATCH v3 6/8] rebase: add --update-refs option","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-28T13:25:57Z","receivedAt":"2022-06-28T13:28:17Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nWhen working on a large feature, it can be helpful to break that feature\ninto multiple smaller parts that become reviewed in sequence. During\ndevelopment or during review, a change to one part of the feature could\naffect multiple of these parts. An interactive rebase can help adjust\nthe multi-part \"story\" of the branch.\n\nHowever, if there are branches tracking the different parts of the\nfeature, then rebasing the entire list of commits can create commits not\nreachable from those \"sub branches\". It can take a manual step to update\nthose branches.\n\nAdd a new --update-refs option to 'git rebase -i' that adds 'update-ref\n' steps to the todo file whenever a commit that is being rebased is\ndecorated with that . At the very end, the rebase process updates\nall of the listed refs to the values stored during the rebase operation.\n\nBe sure to iterate after any squashing or fixups are placed. Update the\nbranch only after those squashes and fixups are complete. This allows a\n--fixup commit at the tip of the feature to apply correctly to the sub\nbranch, even if it is fixing up the most-recent commit in that part.\n\nOne potential problem here is that refs decorating commits that are\nalready marked as \"fixup!\" or \"squash!\" will not be included in this\nlist. Generally, the reordering of the \"fixup!\" and \"squash!\" is likely\nto change the relative order of these refs, so it is not recommended.\nThe workflow here is intended to allow these kinds of commits at the tip\nof the rebased branch while the other sub branches come along for the\nride without intervention.\n\nThis change update the documentation and builtin to accept the\n--update-refs option as well as updating the todo file with the\n'update-ref' commands. Tests are added to ensure that these todo\ncommands are added in the correct locations.\n\nA future change will update the behavior to actually update the refs\nat the end of the rebase sequence.\n\nSigned-off-by: Derrick Stolee \n---\n Documentation/git-rebase.txt | 8 +++\n builtin/rebase.c | 5 ++\n sequencer.c | 107 ++++++++++++++++++++++++++++++++++\n sequencer.h | 1 +\n t/t2407-worktree-heads.sh | 17 ++++++\n t/t3404-rebase-interactive.sh | 70 ++++++++++++++++++++++\n 6 files changed, 208 insertions(+)\n\ndiff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt\nindex 262fb01aec0..e7611b4089c 100644\n--- a/Documentation/git-rebase.txt\n+++ b/Documentation/git-rebase.txt\n@@ -609,6 +609,13 @@ provided. Otherwise an explicit `--no-reschedule-failed-exec` at the\n start would be overridden by the presence of\n `rebase.rescheduleFailedExec=true` configuration.\n \n+--update-refs::\n+--no-update-refs::\n+\tAutomatically force-update any branches that point to commits that\n+\tare being rebased. Any branches that are checked out in a worktree\n+\tor point to a `squash! ...` or `fixup! ...` commit are not updated\n+\tin this way.\n+\n INCOMPATIBLE OPTIONS\n --------------------\n \n@@ -632,6 +639,7 @@ are incompatible with the following options:\n * --empty=\n * --reapply-cherry-picks\n * --edit-todo\n+ * --update-refs\n * --root when used in combination with --onto\n \n In addition, the following pairs of options are incompatible:\ndiff --git a/builtin/rebase.c b/builtin/rebase.c\nindex 7ab50cda2ad..56d82a52106 100644\n--- a/builtin/rebase.c\n+++ b/builtin/rebase.c\n@@ -102,6 +102,7 @@ struct rebase_options {\n \tint reschedule_failed_exec;\n \tint reapply_cherry_picks;\n \tint fork_point;\n+\tint update_refs;\n };\n \n #define REBASE_OPTIONS_INIT {\t\t\t \t\\\n@@ -298,6 +299,7 @@ static int do_interactive_rebase(struct rebase_options *opts, unsigned flags)\n \t\tret = complete_action(the_repository, &replay, flags,\n \t\t\tshortrevisions, opts->onto_name, opts->onto,\n \t\t\t&opts->orig_head, &commands, opts->autosquash,\n+\t\t\topts->update_refs,\n \t\t\t&todo_list);\n \t}\n \n@@ -1124,6 +1126,9 @@ int cmd_rebase(int argc, const char **argv, const char *prefix)\n \t\tOPT_BOOL(0, \"autosquash\", &options.autosquash,\n \t\t\t N_(\"move commits that begin with \"\n \t\t\t \"squash!/fixup! under -i\")),\n+\t\tOPT_BOOL(0, \"update-refs\", &options.update_refs,\n+\t\t\t N_(\"update local refs that point to commits \"\n+\t\t\t \"that are being rebased\")),\n \t\t{ OPTION_STRING, 'S', \"gpg-sign\", &gpg_sign, N_(\"key-id\"),\n \t\t\tN_(\"GPG-sign commits\"),\n \t\t\tPARSE_OPT_OPTARG, NULL, (intptr_t) \"\" },\ndiff --git a/sequencer.c b/sequencer.c\nindex 0b61835d441..915d87a0336 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -35,6 +35,7 @@\n #include \"commit-reach.h\"\n #include \"rebase-interactive.h\"\n #include \"reset.h\"\n+#include \"branch.h\"\n \n #define GIT_REFLOG_ACTION \"GIT_REFLOG_ACTION\"\n \n@@ -5615,10 +5616,113 @@ static int skip_unnecessary_picks(struct repository *r,\n \treturn 0;\n }\n \n+struct todo_add_branch_context {\n+\tstruct todo_item *items;\n+\tsize_t items_nr;\n+\tsize_t items_alloc;\n+\tstruct strbuf *buf;\n+\tstruct commit *commit;\n+\tstruct string_list refs_to_oids;\n+};\n+\n+static int add_decorations_to_list(const struct commit *commit,\n+\t\t\t\t struct todo_add_branch_context *ctx)\n+{\n+\tconst struct name_decoration *decoration = get_name_decoration(&commit->object);\n+\n+\twhile (decoration) {\n+\t\tstruct todo_item *item;\n+\t\tconst char *path;\n+\t\tsize_t base_offset = ctx->buf->len;\n+\n+\t\tALLOC_GROW(ctx->items,\n+\t\t\tctx->items_nr + 1,\n+\t\t\tctx->items_alloc);\n+\t\titem = &ctx->items[ctx->items_nr];\n+\t\tmemset(item, 0, sizeof(*item));\n+\n+\t\t/* If the branch is checked out, then leave a comment instead. */\n+\t\tif ((path = branch_checked_out(decoration->name))) {\n+\t\t\titem->command = TODO_COMMENT;\n+\t\t\tstrbuf_addf(ctx->buf, \"# Ref %s checked out at '%s'\\n\",\n+\t\t\t\t decoration->name, path);\n+\t\t} else {\n+\t\t\tstruct string_list_item *sti;\n+\t\t\titem->command = TODO_UPDATE_REF;\n+\t\t\tstrbuf_addf(ctx->buf, \"%s\\n\", decoration->name);\n+\n+\t\t\tsti = string_list_append(&ctx->refs_to_oids,\n+\t\t\t\t\t\t decoration->name);\n+\t\t\tsti->util = oiddup(the_hash_algo->null_oid);\n+\t\t}\n+\n+\t\titem->offset_in_buf = base_offset;\n+\t\titem->arg_offset = base_offset;\n+\t\titem->arg_len = ctx->buf->len - base_offset;\n+\t\tctx->items_nr++;\n+\n+\t\tdecoration = decoration->next;\n+\t}\n+\n+\treturn 0;\n+}\n+\n+/*\n+ * For each 'pick' command, find out if the commit has a decoration in\n+ * refs/heads/. If so, then add a 'label for-update-refs/' command.\n+ */\n+static int todo_list_add_update_ref_commands(struct todo_list *todo_list)\n+{\n+\tint i;\n+\tstatic struct string_list decorate_refs_exclude = STRING_LIST_INIT_NODUP;\n+\tstatic struct string_list decorate_refs_exclude_config = STRING_LIST_INIT_NODUP;\n+\tstatic struct string_list decorate_refs_include = STRING_LIST_INIT_NODUP;\n+\tstruct decoration_filter decoration_filter = {\n+\t\t.include_ref_pattern = &decorate_refs_include,\n+\t\t.exclude_ref_pattern = &decorate_refs_exclude,\n+\t\t.exclude_ref_config_pattern = &decorate_refs_exclude_config,\n+\t};\n+\tstruct todo_add_branch_context ctx = {\n+\t\t.buf = &todo_list->buf,\n+\t\t.refs_to_oids = STRING_LIST_INIT_DUP,\n+\t};\n+\n+\tctx.items_alloc = 2 * todo_list->nr + 1;\n+\tALLOC_ARRAY(ctx.items, ctx.items_alloc);\n+\n+\tstring_list_append(&decorate_refs_include, \"refs/heads/\");\n+\tload_ref_decorations(&decoration_filter, 0);\n+\n+\tfor (i = 0; i < todo_list->nr; ) {\n+\t\tstruct todo_item *item = &todo_list->items[i];\n+\n+\t\t/* insert ith item into new list */\n+\t\tALLOC_GROW(ctx.items,\n+\t\t\t ctx.items_nr + 1,\n+\t\t\t ctx.items_alloc);\n+\n+\t\tctx.items[ctx.items_nr++] = todo_list->items[i++];\n+\n+\t\tif (item->commit) {\n+\t\t\tctx.commit = item->commit;\n+\t\t\tadd_decorations_to_list(item->commit, &ctx);\n+\t\t}\n+\t}\n+\n+\tstring_list_clear(&ctx.refs_to_oids, 1);\n+\tfree(todo_list->items);\n+\ttodo_list->items = ctx.items;\n+\ttodo_list->nr = ctx.items_nr;\n+\ttodo_list->alloc = ctx.items_alloc;\n+\n+\treturn 0;\n+}\n+\n int complete_action(struct repository *r, struct replay_opts *opts, unsigned flags,\n \t\t const char *shortrevisions, const char *onto_name,\n \t\t struct commit *onto, const struct object_id *orig_head,\n \t\t struct string_list *commands, unsigned autosquash,\n+\t\t unsigned update_refs,\n \t\t struct todo_list *todo_list)\n {\n \tchar shortonto[GIT_MAX_HEXSZ + 1];\n@@ -5637,6 +5741,9 @@ int complete_action(struct repository *r, struct replay_opts *opts, unsigned fla\n \t\titem->arg_len = item->arg_offset = item->flags = item->offset_in_buf = 0;\n \t}\n \n+\tif (update_refs && todo_list_add_update_ref_commands(todo_list))\n+\t\treturn -1;\n+\n \tif (autosquash && todo_list_rearrange_squash(todo_list))\n \t\treturn -1;\n \ndiff --git a/sequencer.h b/sequencer.h\nindex 2cf5c1b9a38..e671d7e0743 100644\n--- a/sequencer.h\n+++ b/sequencer.h\n@@ -167,6 +167,7 @@ int complete_action(struct repository *r, struct replay_opts *opts, unsigned fla\n \t\t const char *shortrevisions, const char *onto_name,\n \t\t struct commit *onto, const struct object_id *orig_head,\n \t\t struct string_list *commands, unsigned autosquash,\n+\t\t unsigned update_refs,\n \t\t struct todo_list *todo_list);\n int todo_list_rearrange_squash(struct todo_list *todo_list);\n \ndiff --git a/t/t2407-worktree-heads.sh b/t/t2407-worktree-heads.sh\nindex f1e9e172a0c..203997d92c6 100755\n--- a/t/t2407-worktree-heads.sh\n+++ b/t/t2407-worktree-heads.sh\n@@ -151,4 +151,21 @@ test_expect_success 'refuse to overwrite when in error states' '\n \tdone\n '\n \n+. \"$TEST_DIRECTORY\"/lib-rebase.sh\n+\n+test_expect_success !SANITIZE_LEAK 'refuse to overwrite during rebase with --update-refs' '\n+\tgit commit --fixup HEAD~2 --allow-empty &&\n+\t(\n+\t\tset_cat_todo_editor &&\n+\t\ttest_must_fail git rebase -i --update-refs HEAD~3 >todo &&\n+\t\t! grep \"update-refs\" todo\n+\t) &&\n+\tgit branch -f allow-update HEAD~2 &&\n+\t(\n+\t\tset_cat_todo_editor &&\n+\t\ttest_must_fail git rebase -i --update-refs HEAD~3 >todo &&\n+\t\tgrep \"update-ref refs/heads/allow-update\" todo\n+\t)\n+'\n+\n test_done\ndiff --git a/t/t3404-rebase-interactive.sh b/t/t3404-rebase-interactive.sh\nindex f31afd4a547..3cd20733bc8 100755\n--- a/t/t3404-rebase-interactive.sh\n+++ b/t/t3404-rebase-interactive.sh\n@@ -1743,6 +1743,76 @@ test_expect_success 'ORIG_HEAD is updated correctly' '\n \ttest_cmp_rev ORIG_HEAD test-orig-head@{1}\n '\n \n+test_expect_success '--update-refs adds label and update-ref commands' '\n+\tgit checkout -b update-refs no-conflict-branch &&\n+\tgit branch -f base HEAD~4 &&\n+\tgit branch -f first HEAD~3 &&\n+\tgit branch -f second HEAD~3 &&\n+\tgit branch -f third HEAD~1 &&\n+\tgit commit --allow-empty --fixup=third &&\n+\tgit branch -f shared-tip &&\n+\t(\n+\t\tset_cat_todo_editor &&\n+\n+\t\tcat >expect <<-EOF &&\n+\t\tpick $(git log -1 --format=%h J) J\n+\t\tupdate-ref refs/heads/second\n+\t\tupdate-ref refs/heads/first\n+\t\tpick $(git log -1 --format=%h K) K\n+\t\tpick $(git log -1 --format=%h L) L\n+\t\tfixup $(git log -1 --format=%h update-refs) fixup! L # empty\n+\t\tupdate-ref refs/heads/third\n+\t\tpick $(git log -1 --format=%h M) M\n+\t\tupdate-ref refs/heads/no-conflict-branch\n+\t\tupdate-ref refs/heads/shared-tip\n+\t\tEOF\n+\n+\t\ttest_must_fail git rebase -i --autosquash --update-refs primary >todo &&\n+\t\ttest_cmp expect todo\n+\t)\n+'\n+\n+test_expect_success '--update-refs adds commands with --rebase-merges' '\n+\tgit checkout -b update-refs-with-merge no-conflict-branch &&\n+\tgit branch -f base HEAD~4 &&\n+\tgit branch -f first HEAD~3 &&\n+\tgit branch -f second HEAD~3 &&\n+\tgit branch -f third HEAD~1 &&\n+\tgit merge -m merge branch2 &&\n+\tgit branch -f merge-branch &&\n+\tgit commit --fixup=third --allow-empty &&\n+\t(\n+\t\tset_cat_todo_editor &&\n+\n+\t\tcat >expect <<-EOF &&\n+\t\tlabel onto\n+\t\treset onto\n+\t\tpick $(git log -1 --format=%h branch2~1) F\n+\t\tpick $(git log -1 --format=%h branch2) I\n+\t\tupdate-ref refs/heads/branch2\n+\t\tlabel merge\n+\t\treset onto\n+\t\tpick $(git log -1 --format=%h refs/heads/second) J\n+\t\tupdate-ref refs/heads/second\n+\t\tupdate-ref refs/heads/first\n+\t\tpick $(git log -1 --format=%h refs/heads/third~1) K\n+\t\tpick $(git log -1 --format=%h refs/heads/third) L\n+\t\tfixup $(git log -1 --format=%h update-refs-with-merge) fixup! L # empty\n+\t\tupdate-ref refs/heads/third\n+\t\tpick $(git log -1 --format=%h HEAD~2) M\n+\t\tupdate-ref refs/heads/no-conflict-branch\n+\t\tmerge -C $(git log -1 --format=%h HEAD~1) merge # merge\n+\t\tupdate-ref refs/heads/merge-branch\n+\t\tEOF\n+\n+\t\ttest_must_fail git rebase -i --autosquash \\\n+\t\t\t\t --rebase-merges=rebase-cousins \\\n+\t\t\t\t --update-refs primary >todo &&\n+\n+\t\ttest_cmp expect todo\n+\t)\n+'\n+\n # This must be the last test in this file\n test_expect_success '$EDITOR and friends are unchanged' '\n \ttest_editor_unchanged\n-- \ngitgitgadget\n\n"},{"id":"458038","messageId":"pull.1247.v3.git.1656422759.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v2.git.1654634569.gitgitgadget@gmail.com","subject":"[PATCH v3 0/8] rebase: update branches in multi-part topic","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-28T13:25:51Z","receivedAt":"2022-06-28T13:28:20Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"(Update: This is now based on ds/branch-checked-out.)\n\nThis is a feature I've wanted for quite a while. When working on the sparse\nindex topic, I created a long RFC that actually broke into three topics for\nfull review upstream. These topics were sequential, so any feedback on an\nearlier one required updates to the later ones. I would work on the full\nfeature and use interactive rebase to update the full list of commits.\nHowever, I would need to update the branches pointing to those sub-topics.\n\nThis series adds a new --update-refs option to 'git rebase' (along with a\nrebase.updateRefs config option) that adds 'update-ref' commands into the\nTODO list. This is powered by the commit decoration machinery.\n\nAs an example, here is my in-progress bundle URI RFC split into subtopics as\nthey appear during the TODO list of a git rebase -i --update-refs:\n\npick 2d966282ff3 docs: document bundle URI standard\npick 31396e9171a remote-curl: add 'get' capability\npick 54c6ab70f67 bundle-uri: create basic file-copy logic\npick 96cb2e35af1 bundle-uri: add support for http(s):// and file://\npick 6adaf842684 fetch: add --bundle-uri option\npick 6c5840ed77e fetch: add 'refs/bundle/' to log.excludeDecoration\nupdate-ref refs/heads/bundle-redo/fetch\n\npick 1e3f6546632 clone: add --bundle-uri option\npick 9e4a6fe9b68 clone: --bundle-uri cannot be combined with --depth\nupdate-ref refs/heads/bundle-redo/clone\n\npick 5451cb6599c bundle-uri: create bundle_list struct and helpers\npick 3029c3aca15 bundle-uri: create base key-value pair parsing\npick a8b2de79ce8 bundle-uri: create \"key=value\" line parsing\npick 92625a47673 bundle-uri: unit test \"key=value\" parsing\npick a8616af4dc2 bundle-uri: limit recursion depth for bundle lists\npick 9d6809a8d53 bundle-uri: parse bundle list in config format\npick 287a732b54c bundle-uri: fetch a list of bundles\nupdate-ref refs/heads/bundle-redo/list\n\npick b09f8226185 protocol v2: add server-side \"bundle-uri\" skeleton\npick 520204dcd1c bundle-uri client: add minimal NOOP client\npick 62e8b457b48 bundle-uri client: add \"git ls-remote-bundle-uri\"\npick 00eae925043 bundle-uri: serve URI advertisement from bundle.* config\npick 4277440a250 bundle-uri client: add boolean transfer.bundleURI setting\npick caf4599a81d bundle-uri: allow relative URLs in bundle lists\npick df255000b7e bundle-uri: download bundles from an advertised list\npick d71beabf199 clone: unbundle the advertised bundles\npick c9578391976 t5601: basic bundle URI tests\n# Ref refs/heads/bundle-redo/rfc-3 checked out at '/home/stolee/_git/git-bundles'\n\nupdate-ref refs/heads/bundle-redo/advertise\n\n\nHere is an outline of the series:\n\n * Patch 1 updates some tests for branch_checked_out() for the 'apply'\n backend.\n * Patch 2 updates branch_checked_out() to parse the\n rebase-merge/update-refs file to block concurrent ref updates and\n checkouts on branches selected by --update-refs.\n * Patch 3 updates the todo list documentation to remove some unnecessary\n dots in the 'merge' command. This makes it consistent with the 'fixup'\n command before we document the 'update-ref' command.\n * Patch 4 updates the definition of todo_command_info to use enum values as\n array indices.\n * Patches 5-7 implement the --update-refs logic itself.\n * Patch 8 adds the rebase.updateRefs config option similar to\n rebase.autoSquash.\n\n\nUpdates in v3\n=============\n\n * The branch_checked_out() API was extracted to its own topic and is now\n the ds/branch-checked-out branch. This series is now based on that one.\n * The for_each_decoration() API was removed, since it became trivial once\n it did not take a commit directly.\n * The branch_checked_out() tests did not verify the rebase-apply data (for\n the apply backend), so that is fixed.\n * Instead of using the 'label' command and a final 'update-refs' command in\n the todo list, use a new 'update-ref ' command. This command updates the\n rebase-merge/update-refs file with the OID of HEAD at these steps. At the\n very end of the rebase sequence, those refs are updated to the stored OID\n values (assuming that they were not removed by the user, in which case we\n notice that the OID is the null OID and we do nothing).\n * New tests are added.\n * The todo-list comment documentation has some new formatting updates, but\n also includes a description of 'update-refs' in this version.\n\n\nUpdates in v2\n=============\n\nAs recommended by the excellent feedback, I have removed the 'exec' commands\nin favor of the 'label' commands and a new 'update-refs' command at the very\nend. This way, there is only one step that updates all of the refs at the\nend instead of updating refs during the rebase. If a user runs 'git rebase\n--abort' in the middle, then their refs are still where they need to be.\n\nBased on some of the discussion, it seemed like one way to do this would be\nto have an 'update-ref ' command that would take the place of these 'label'\ncommands. However, this would require two things that make it a bit awkward:\n\n 1. We would need to replicate the storage of those positions during the\n rebase. 'label' already does this pretty well. I've added the\n \"for-update-refs/\" label to help here.\n 2. If we want to close out all of the refs as the rebase is finishing, then\n that \"step\" becomes invisible to the user (and a bit more complicated to\n insert). Thus, the 'update-refs' step performs this action. If the user\n wants to do things after that step, then they can do so by editing the\n TODO list.\n\nOther updates:\n\n * The 'keep_decorations' parameter was renamed to 'update_refs'.\n * I added tests for --rebase-merges=rebase-cousins to show how these labels\n interact with other labels and merge commands.\n * I changed the order of the insertion of these update-refs labels to be\n before the fixups are rearranged. This fixes a bug where the tip commit\n is a fixup! so its decorations are never inspected (and they would be in\n the wrong place even if they were). The fixup! commands are properly\n inserted between a pick and its following label command. Tests\n demonstrate this is correct.\n * Numerous style choices are updated based on feedback.\n\nThank you for all of the detailed review and ideas in this space. I\nappreciate any more ideas that can make this feature as effective as it can\nbe.\n\nThanks, -Stolee\n\nDerrick Stolee (8):\n t2407: test branches currently using apply backend\n branch: consider refs under 'update-refs'\n rebase-interactive: update 'merge' description\n sequencer: define array with enum values\n sequencer: add update-ref command\n rebase: add --update-refs option\n rebase: update refs from 'update-ref' commands\n rebase: add rebase.updateRefs config option\n\n Documentation/config/rebase.txt | 3 +\n Documentation/git-rebase.txt | 11 ++\n branch.c | 13 ++\n builtin/rebase.c | 10 ++\n rebase-interactive.c | 9 +-\n sequencer.c | 303 ++++++++++++++++++++++++++++++--\n sequencer.h | 11 ++\n t/t2407-worktree-heads.sh | 52 +++++-\n t/t3404-rebase-interactive.sh | 107 +++++++++++\n 9 files changed, 496 insertions(+), 23 deletions(-)\n\n\nbase-commit: 9bef0b1e6ec371e786c2fba3edcc06ad040a536c\nPublished-As: https://github.com/gitgitgadget/git/releases/tag/pr-1247%2Fderrickstolee%2Frebase-keep-decorations-v3\nFetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1247/derrickstolee/rebase-keep-decorations-v3\nPull-Request: https://github.com/gitgitgadget/git/pull/1247\n\nRange-diff vs v2:\n\n 1: 4f9f3487641 < -: ----------- log-tree: create for_each_decoration()\n 2: 5f54766e103 < -: ----------- branch: add branch_checked_out() helper\n -: ----------- > 1: fbaedc7f1f0 t2407: test branches currently using apply backend\n -: ----------- > 2: 2bc647b6fcd branch: consider refs under 'update-refs'\n -: ----------- > 3: 669f4abd59e rebase-interactive: update 'merge' description\n 3: 9f261c7df2c = 4: 6528a50343f sequencer: define array with enum values\n 4: 842b2186d25 ! 5: e95ad41d355 sequencer: add update-refs command\n @@ Metadata\n Author: Derrick Stolee \n \n ## Commit message ##\n - sequencer: add update-refs command\n + sequencer: add update-ref command\n \n - Add the boilerplat for an \"update-refs\" command in the sequencer. This\n - connects to the current no-op do_update_refs() which will be filled in\n + Add the boilerplate for an \"update-ref\" command in the sequencer. This\n + connects to the current no-op do_update_ref() which will be filled in\n after more connections are created.\n \n + The syntax in the todo list will be \"update-ref \" to signal\n + that we should store the current commit as the value for updating\n + at the end of the rebase.\n +\n Signed-off-by: Derrick Stolee \n \n + ## rebase-interactive.c ##\n +@@ rebase-interactive.c: void append_todo_help(int command_count,\n + \" create a merge commit using the original merge commit's\\n\"\n + \" message (or the oneline, if no original merge commit was\\n\"\n + \" specified); use -c to reword the commit message\\n\"\n ++\"u, update-ref = track a placeholder for the to be updated\\n\"\n ++\" to this position in the new commits. The is\\n\"\n ++\" updated at the end of the rebase\\n\"\n + \"\\n\"\n + \"These lines can be re-ordered; they are executed from top to bottom.\\n\");\n + \tunsigned edit_todo = !(shortrevisions && shortonto);\n +\n ## sequencer.c ##\n @@ sequencer.c: static struct {\n \t[TODO_LABEL] = { 'l', \"label\" },\n \t[TODO_RESET] = { 't', \"reset\" },\n \t[TODO_MERGE] = { 'm', \"merge\" },\n -+\t[TODO_UPDATE_REFS] = { 'u', \"update-refs\" },\n ++\t[TODO_UPDATE_REF] = { 'u', \"update-ref\" },\n \t[TODO_NOOP] = { 0, \"noop\" },\n \t[TODO_DROP] = { 'd', \"drop\" },\n \t[TODO_COMMENT] = { 0, NULL },\n @@ sequencer.c: static int parse_insn_line(struct repository *r, struct todo_item *item,\n - \tpadding = strspn(bol, \" \\t\");\n - \tbol += padding;\n + \t\t\t command_to_string(item->command));\n \n --\tif (item->command == TODO_NOOP || item->command == TODO_BREAK) {\n -+\tif (item->command == TODO_NOOP ||\n -+\t item->command == TODO_BREAK ||\n -+\t item->command == TODO_UPDATE_REFS) {\n - \t\tif (bol != eol)\n - \t\t\treturn error(_(\"%s does not accept arguments: '%s'\"),\n - \t\t\t\t command_to_string(item->command), bol);\n + \tif (item->command == TODO_EXEC || item->command == TODO_LABEL ||\n +-\t item->command == TODO_RESET) {\n ++\t item->command == TODO_RESET || item->command == TODO_UPDATE_REF) {\n + \t\titem->commit = NULL;\n + \t\titem->arg_offset = bol - buf;\n + \t\titem->arg_len = (int)(eol - bol);\n @@ sequencer.c: leave_merge:\n \treturn ret;\n }\n \n -+static int do_update_refs(struct repository *r)\n ++static int do_update_ref(struct repository *r, const char *ref_name)\n +{\n +\treturn 0;\n +}\n @@ sequencer.c: static int pick_commits(struct repository *r,\n \t\t\t\treturn error_with_patch(r, item->commit,\n \t\t\t\t\t\t\targ, item->arg_len,\n \t\t\t\t\t\t\topts, res, 0);\n -+\t\t} else if (item->command == TODO_UPDATE_REFS) {\n -+\t\t\tif ((res = do_update_refs(r)))\n ++\t\t} else if (item->command == TODO_UPDATE_REF) {\n ++\t\t\tstruct strbuf ref = STRBUF_INIT;\n ++\t\t\tstrbuf_add(&ref, arg, item->arg_len);\n ++\t\t\tif ((res = do_update_ref(r, ref.buf)))\n +\t\t\t\treschedule = 1;\n ++\t\t\tstrbuf_release(&ref);\n \t\t} else if (!is_noop(item->command))\n \t\t\treturn error(_(\"unknown command %d\"), item->command);\n \n @@ sequencer.h: enum todo_command {\n \tTODO_LABEL,\n \tTODO_RESET,\n \tTODO_MERGE,\n -+\tTODO_UPDATE_REFS,\n ++\tTODO_UPDATE_REF,\n \t/* commands that do nothing but are counted for reporting progress */\n \tTODO_NOOP,\n \tTODO_DROP,\n 5: 0a4c110127b ! 6: 918b398d6a2 rebase: add --update-refs option\n @@ Commit message\n reachable from those \"sub branches\". It can take a manual step to update\n those branches.\n \n - Add a new --update-refs option to 'git rebase -i' that adds 'label\n - for-update-refs/*' steps to the todo file whenever a commit that is\n - being rebased is decorated with that . At the very end, the\n - 'update-refs' step is added to update all of the branches referenced by\n - the 'label' steps. This allows the user to rebase a long list of commits\n - in a multi-part feature and keep all of their pointers to those parts.\n + Add a new --update-refs option to 'git rebase -i' that adds 'update-ref\n + ' steps to the todo file whenever a commit that is being rebased is\n + decorated with that . At the very end, the rebase process updates\n + all of the listed refs to the values stored during the rebase operation.\n \n - NOTE: This change only introduce the --update-refs option and implements\n - the changes to the todo file. It does _not_ yet implement the action\n - taken by the 'update-refs' todo step, which will be implemented and\n - tested in a later change.\n -\n - Use the new for_each_decoration() while iterating over the existing todo\n - list. Be sure to iterate after any squashing or fixups are placed.\n - Update the branch only after those squashes and fixups are complete.\n - This allows a --fixup commit at the tip of the feature to apply\n - correctly to the sub branch, even if it is fixing up the most-recent\n - commit in that part.\n + Be sure to iterate after any squashing or fixups are placed. Update the\n + branch only after those squashes and fixups are complete. This allows a\n + --fixup commit at the tip of the feature to apply correctly to the sub\n + branch, even if it is fixing up the most-recent commit in that part.\n \n One potential problem here is that refs decorating commits that are\n already marked as \"fixup!\" or \"squash!\" will not be included in this\n @@ Commit message\n of the rebased branch while the other sub branches come along for the\n ride without intervention.\n \n - Be careful to not attempt updating any branch that is checked out. The\n - most common example is the branch being rebased is checked out and\n - decorates the tip commit. If the user is rebasing commits reachable from\n - a different branch that is checked out in a different worktree, then\n - they may be surprised to not see that ref update. However, it's probably\n - best to not optimize for this scenario and do the safest thing that will\n - result in a successful rebase. A comment is left in the TODO list that\n - signals that these refs are currently checked out.\n + This change update the documentation and builtin to accept the\n + --update-refs option as well as updating the todo file with the\n + 'update-ref' commands. Tests are added to ensure that these todo\n + commands are added in the correct locations.\n +\n + A future change will update the behavior to actually update the refs\n + at the end of the rebase sequence.\n \n Signed-off-by: Derrick Stolee \n \n @@ sequencer.c\n #include \"rebase-interactive.h\"\n #include \"reset.h\"\n +#include \"branch.h\"\n -+#include \"log-tree.h\"\n \n #define GIT_REFLOG_ACTION \"GIT_REFLOG_ACTION\"\n \n @@ sequencer.c: static int skip_unnecessary_picks(struct repository *r,\n +\tsize_t items_alloc;\n +\tstruct strbuf *buf;\n +\tstruct commit *commit;\n ++\tstruct string_list refs_to_oids;\n +};\n +\n -+static int add_branch_for_decoration(const struct name_decoration *d, void *data)\n ++static int add_decorations_to_list(const struct commit *commit,\n ++\t\t\t\t struct todo_add_branch_context *ctx)\n +{\n -+\tstruct todo_add_branch_context *ctx = data;\n -+\tsize_t base_offset = ctx->buf->len;\n -+\tstruct todo_item *item;\n -+\tchar *path;\n ++\tconst struct name_decoration *decoration = get_name_decoration(&commit->object);\n +\n -+\tALLOC_GROW(ctx->items,\n -+\t\t ctx->items_nr + 1,\n -+\t\t ctx->items_alloc);\n -+\titem = &ctx->items[ctx->items_nr];\n -+\tmemset(item, 0, sizeof(*item));\n ++\twhile (decoration) {\n ++\t\tstruct todo_item *item;\n ++\t\tconst char *path;\n ++\t\tsize_t base_offset = ctx->buf->len;\n +\n -+\t/* If the branch is checked out, then leave a comment instead. */\n -+\tif (branch_checked_out(d->name, &path)) {\n -+\t\titem->command = TODO_COMMENT;\n -+\t\tstrbuf_addf(ctx->buf, \"# Ref %s checked out at '%s'\\n\",\n -+\t\t\t d->name, path);\n -+\t\tfree(path);\n -+\t} else {\n -+\t\titem->command = TODO_LABEL;\n -+\t\tstrbuf_addf(ctx->buf, \"for-update-refs/%s\\n\", d->name);\n -+\t}\n ++\t\tALLOC_GROW(ctx->items,\n ++\t\t\tctx->items_nr + 1,\n ++\t\t\tctx->items_alloc);\n ++\t\titem = &ctx->items[ctx->items_nr];\n ++\t\tmemset(item, 0, sizeof(*item));\n ++\n ++\t\t/* If the branch is checked out, then leave a comment instead. */\n ++\t\tif ((path = branch_checked_out(decoration->name))) {\n ++\t\t\titem->command = TODO_COMMENT;\n ++\t\t\tstrbuf_addf(ctx->buf, \"# Ref %s checked out at '%s'\\n\",\n ++\t\t\t\t decoration->name, path);\n ++\t\t} else {\n ++\t\t\tstruct string_list_item *sti;\n ++\t\t\titem->command = TODO_UPDATE_REF;\n ++\t\t\tstrbuf_addf(ctx->buf, \"%s\\n\", decoration->name);\n ++\n ++\t\t\tsti = string_list_append(&ctx->refs_to_oids,\n ++\t\t\t\t\t\t decoration->name);\n ++\t\t\tsti->util = oiddup(the_hash_algo->null_oid);\n ++\t\t}\n ++\n ++\t\titem->offset_in_buf = base_offset;\n ++\t\titem->arg_offset = base_offset;\n ++\t\titem->arg_len = ctx->buf->len - base_offset;\n ++\t\tctx->items_nr++;\n +\n -+\titem->offset_in_buf = base_offset;\n -+\titem->arg_offset = base_offset;\n -+\titem->arg_len = ctx->buf->len - base_offset;\n -+\tctx->items_nr++;\n ++\t\tdecoration = decoration->next;\n ++\t}\n +\n +\treturn 0;\n +}\n @@ sequencer.c: static int skip_unnecessary_picks(struct repository *r,\n +\tstruct decoration_filter decoration_filter = {\n +\t\t.include_ref_pattern = &decorate_refs_include,\n +\t\t.exclude_ref_pattern = &decorate_refs_exclude,\n -+\t\t.exclude_ref_config_pattern = &decorate_refs_exclude_config\n ++\t\t.exclude_ref_config_pattern = &decorate_refs_exclude_config,\n +\t};\n +\tstruct todo_add_branch_context ctx = {\n +\t\t.buf = &todo_list->buf,\n ++\t\t.refs_to_oids = STRING_LIST_INIT_DUP,\n +\t};\n +\n +\tctx.items_alloc = 2 * todo_list->nr + 1;\n @@ sequencer.c: static int skip_unnecessary_picks(struct repository *r,\n +\n +\t\tif (item->commit) {\n +\t\t\tctx.commit = item->commit;\n -+\t\t\tfor_each_decoration(item->commit,\n -+\t\t\t\t\t add_branch_for_decoration,\n -+\t\t\t\t\t &ctx);\n ++\t\t\tadd_decorations_to_list(item->commit, &ctx);\n +\t\t}\n +\t}\n +\n -+\t/* Add the \"update-refs\" step. */\n -+\tALLOC_GROW(ctx.items,\n -+\t\t ctx.items_nr + 1,\n -+\t\t ctx.items_alloc);\n -+\tmemset(&ctx.items[ctx.items_nr], 0, sizeof(struct todo_item));\n -+\tctx.items[ctx.items_nr].command = TODO_UPDATE_REFS;\n -+\tctx.items_nr++;\n -+\n ++\tstring_list_clear(&ctx.refs_to_oids, 1);\n +\tfree(todo_list->items);\n +\ttodo_list->items = ctx.items;\n +\ttodo_list->nr = ctx.items_nr;\n @@ sequencer.h: int complete_action(struct repository *r, struct replay_opts *opts,\n int todo_list_rearrange_squash(struct todo_list *todo_list);\n \n \n + ## t/t2407-worktree-heads.sh ##\n +@@ t/t2407-worktree-heads.sh: test_expect_success 'refuse to overwrite when in error states' '\n + \tdone\n + '\n + \n ++. \"$TEST_DIRECTORY\"/lib-rebase.sh\n ++\n ++test_expect_success !SANITIZE_LEAK 'refuse to overwrite during rebase with --update-refs' '\n ++\tgit commit --fixup HEAD~2 --allow-empty &&\n ++\t(\n ++\t\tset_cat_todo_editor &&\n ++\t\ttest_must_fail git rebase -i --update-refs HEAD~3 >todo &&\n ++\t\t! grep \"update-refs\" todo\n ++\t) &&\n ++\tgit branch -f allow-update HEAD~2 &&\n ++\t(\n ++\t\tset_cat_todo_editor &&\n ++\t\ttest_must_fail git rebase -i --update-refs HEAD~3 >todo &&\n ++\t\tgrep \"update-ref refs/heads/allow-update\" todo\n ++\t)\n ++'\n ++\n + test_done\n +\n ## t/t3404-rebase-interactive.sh ##\n @@ t/t3404-rebase-interactive.sh: test_expect_success 'ORIG_HEAD is updated correctly' '\n \ttest_cmp_rev ORIG_HEAD test-orig-head@{1}\n @@ t/t3404-rebase-interactive.sh: test_expect_success 'ORIG_HEAD is updated correct\n +\n +\t\tcat >expect <<-EOF &&\n +\t\tpick $(git log -1 --format=%h J) J\n -+\t\tlabel for-update-refs/refs/heads/second\n -+\t\tlabel for-update-refs/refs/heads/first\n ++\t\tupdate-ref refs/heads/second\n ++\t\tupdate-ref refs/heads/first\n +\t\tpick $(git log -1 --format=%h K) K\n +\t\tpick $(git log -1 --format=%h L) L\n +\t\tfixup $(git log -1 --format=%h update-refs) fixup! L # empty\n -+\t\tlabel for-update-refs/refs/heads/third\n ++\t\tupdate-ref refs/heads/third\n +\t\tpick $(git log -1 --format=%h M) M\n -+\t\tlabel for-update-refs/refs/heads/no-conflict-branch\n -+\t\tlabel for-update-refs/refs/heads/shared-tip\n -+\t\tupdate-refs\n ++\t\tupdate-ref refs/heads/no-conflict-branch\n ++\t\tupdate-ref refs/heads/shared-tip\n +\t\tEOF\n +\n +\t\ttest_must_fail git rebase -i --autosquash --update-refs primary >todo &&\n @@ t/t3404-rebase-interactive.sh: test_expect_success 'ORIG_HEAD is updated correct\n +\t\treset onto\n +\t\tpick $(git log -1 --format=%h branch2~1) F\n +\t\tpick $(git log -1 --format=%h branch2) I\n -+\t\tlabel for-update-refs/refs/heads/branch2\n ++\t\tupdate-ref refs/heads/branch2\n +\t\tlabel merge\n +\t\treset onto\n +\t\tpick $(git log -1 --format=%h refs/heads/second) J\n -+\t\tlabel for-update-refs/refs/heads/second\n -+\t\tlabel for-update-refs/refs/heads/first\n ++\t\tupdate-ref refs/heads/second\n ++\t\tupdate-ref refs/heads/first\n +\t\tpick $(git log -1 --format=%h refs/heads/third~1) K\n +\t\tpick $(git log -1 --format=%h refs/heads/third) L\n +\t\tfixup $(git log -1 --format=%h update-refs-with-merge) fixup! L # empty\n -+\t\tlabel for-update-refs/refs/heads/third\n ++\t\tupdate-ref refs/heads/third\n +\t\tpick $(git log -1 --format=%h HEAD~2) M\n -+\t\tlabel for-update-refs/refs/heads/no-conflict-branch\n ++\t\tupdate-ref refs/heads/no-conflict-branch\n +\t\tmerge -C $(git log -1 --format=%h HEAD~1) merge # merge\n -+\t\tlabel for-update-refs/refs/heads/merge-branch\n -+\t\tupdate-refs\n ++\t\tupdate-ref refs/heads/merge-branch\n +\t\tEOF\n +\n +\t\ttest_must_fail git rebase -i --autosquash \\\n 6: 68f8e51b19c ! 7: 72e0481b643 sequencer: implement 'update-refs' command\n @@ Metadata\n Author: Derrick Stolee \n \n ## Commit message ##\n - sequencer: implement 'update-refs' command\n + rebase: update refs from 'update-ref' commands\n \n - The previous change allowed 'git rebase --update-refs' to create 'label'\n - commands for each branch among the commits being rewritten and add an\n - 'update-refs' command at the end of the todo list. Now, teach Git to\n - update the refs during that final 'update-refs' command.\n + The previous change introduced the 'git rebase --update-refs' option\n + which added 'update-ref ' commands to the todo list of an\n + interactive rebase.\n \n - We need to create an array of new and old OIDs for each ref by iterating\n - over the refs/rewritten/for-update-refs/ namespace. We cannot update the\n - refs in-place since this will confuse the refs iterator.\n + Teach Git to record the HEAD position when reaching these 'update-ref'\n + commands. The ref/OID pair is stored in the\n + $GIT_DIR/rebase-merge/update-refs file. A previous change parsed this\n + file to avoid having other processes updating the refs in that file\n + while the rebase is in progress.\n +\n + Not only do we update the file when the sequencer reaches these\n + 'update-ref' commands, we then update the refs themselves at the end of\n + the rebase sequence. If the rebase is aborted before this final step,\n + then the refs are not updated.\n \n Signed-off-by: Derrick Stolee \n \n ## sequencer.c ##\n +@@\n + #include \"rebase-interactive.h\"\n + #include \"reset.h\"\n + #include \"branch.h\"\n ++#include \"log-tree.h\"\n + \n + #define GIT_REFLOG_ACTION \"GIT_REFLOG_ACTION\"\n + \n +@@ sequencer.c: static GIT_PATH_FUNC(rebase_path_squash_onto, \"rebase-merge/squash-onto\")\n + */\n + static GIT_PATH_FUNC(rebase_path_refs_to_delete, \"rebase-merge/refs-to-delete\")\n + \n ++/*\n ++ * The update-refs file stores a list of refs that will be updated at the end\n ++ * of the rebase sequence. The 'update-ref ' commands in the todo file\n ++ * update the OIDs for the refs in this file, but the refs are not updated\n ++ * until the end of the rebase sequence.\n ++ */\n ++static GIT_PATH_FUNC(rebase_path_update_refs, \"rebase-merge/update-refs\")\n ++\n + /*\n + * The following files are written by git-rebase just after parsing the\n + * command-line.\n @@ sequencer.c: leave_merge:\n \treturn ret;\n }\n \n --static int do_update_refs(struct repository *r)\n -+struct update_refs_context {\n -+\tstruct ref_store *refs;\n -+\tchar **ref_names;\n -+\tstruct object_id *old;\n -+\tstruct object_id *new;\n -+\tsize_t nr;\n -+\tsize_t alloc;\n -+};\n -+\n -+static int add_ref_to_context(const char *refname,\n -+\t\t\t const struct object_id *oid,\n -+\t\t\t int flags,\n -+\t\t\t void *data)\n +-static int do_update_ref(struct repository *r, const char *ref_name)\n ++static int write_update_refs_state(struct string_list *refs_to_oids)\n ++{\n ++\tint result = 0;\n ++\tFILE *fp = NULL;\n ++\tstruct string_list_item *item;\n ++\tchar *path = xstrdup(rebase_path_update_refs());\n ++\n ++\tif (safe_create_leading_directories(path)) {\n ++\t\tresult = error(_(\"unable to create leading directories of %s\"),\n ++\t\t\t path);\n ++\t\tgoto cleanup;\n ++\t}\n ++\n ++\tfp = fopen(path, \"w\");\n ++\tif (!fp) {\n ++\t\tresult = error_errno(_(\"could not open '%s' for writing\"), path);\n ++\t\tgoto cleanup;\n ++\t}\n ++\n ++\tfor_each_string_list_item(item, refs_to_oids)\n ++\t\tfprintf(fp, \"%s\\n%s\\n\", item->string, oid_to_hex(item->util));\n ++\n ++cleanup:\n ++\tif (fp)\n ++\t\tfclose(fp);\n ++\treturn result;\n ++}\n ++\n ++static int do_update_ref(struct repository *r, const char *refname)\n {\n -+\tint f = 0;\n -+\tconst char *name;\n -+\tstruct update_refs_context *ctx = data;\n ++\tstruct string_list_item *item;\n ++\tstruct string_list list = STRING_LIST_INIT_DUP;\n ++\tint found = 0;\n ++\n ++\tsequencer_get_update_refs_state(r->gitdir, &list);\n +\n -+\tALLOC_GROW(ctx->ref_names, ctx->nr + 1, ctx->alloc);\n -+\tALLOC_GROW(ctx->old, ctx->nr + 1, ctx->alloc);\n -+\tALLOC_GROW(ctx->new, ctx->nr + 1, ctx->alloc);\n ++\tfor_each_string_list_item(item, &list) {\n ++\t\tif (!strcmp(item->string, refname)) {\n ++\t\t\tstruct object_id oid;\n ++\t\t\tfree(item->util);\n ++\t\t\tfound = 1;\n +\n -+\tif (!skip_prefix(refname, \"refs/rewritten/for-update-refs/\", &name))\n -+\t\treturn 1;\n ++\t\t\tif (!read_ref(\"HEAD\", &oid)) {\n ++\t\t\t\titem->util = oiddup(&oid);\n ++\t\t\t\tbreak;\n ++\t\t\t}\n ++\t\t}\n ++\t}\n +\n -+\tctx->ref_names[ctx->nr] = xstrdup(name);\n -+\toidcpy(&ctx->new[ctx->nr], oid);\n -+\tif (!refs_resolve_ref_unsafe(ctx->refs, name, 0,\n -+\t\t\t\t &ctx->old[ctx->nr], &f))\n -+\t\treturn 1;\n ++\tif (!found) {\n ++\t\tstruct object_id oid;\n ++\t\titem = string_list_append(&list, refname);\n +\n -+\tctx->nr++;\n ++\t\tif (!read_ref(\"HEAD\", &oid))\n ++\t\t\titem->util = oiddup(&oid);\n ++\t\telse\n ++\t\t\titem->util = oiddup(the_hash_algo->null_oid);\n ++\t}\n ++\n ++\twrite_update_refs_state(&list);\n ++\tstring_list_clear(&list, 1);\n \treturn 0;\n }\n \n +static int do_update_refs(struct repository *r)\n +{\n -+\tint i, res;\n -+\tstruct update_refs_context ctx = {\n -+\t\t.refs = get_main_ref_store(r),\n -+\t\t.alloc = 16,\n -+\t};\n -+\tALLOC_ARRAY(ctx.ref_names, ctx.alloc);\n -+\tALLOC_ARRAY(ctx.old, ctx.alloc);\n -+\tALLOC_ARRAY(ctx.new, ctx.alloc);\n -+\n -+\tres = refs_for_each_fullref_in(ctx.refs,\n -+\t\t\t\t \"refs/rewritten/for-update-refs/\",\n -+\t\t\t\t add_ref_to_context,\n -+\t\t\t\t &ctx);\n -+\n -+\tfor (i = 0; !res && i < ctx.nr; i++)\n -+\t\tres = refs_update_ref(ctx.refs, \"rewritten during rebase\",\n -+\t\t\t\tctx.ref_names[i],\n -+\t\t\t\t&ctx.new[i], &ctx.old[i],\n ++\tint res = 0;\n ++\tstruct string_list_item *item;\n ++\tstruct string_list refs_to_oids = STRING_LIST_INIT_DUP;\n ++\tstruct ref_store *refs = get_main_ref_store(r);\n ++\n ++\tsequencer_get_update_refs_state(r->gitdir, &refs_to_oids);\n ++\n ++\tfor_each_string_list_item(item, &refs_to_oids) {\n ++\t\tstruct object_id *oid_to = item->util;\n ++\t\tstruct object_id oid_from;\n ++\n ++\t\tif (oideq(oid_to, the_hash_algo->null_oid)) {\n ++\t\t\t/*\n ++\t\t\t * Ref was not updated. User may have deleted the\n ++\t\t\t * 'update-ref' step.\n ++\t\t\t */\n ++\t\t\tcontinue;\n ++\t\t}\n ++\n ++\t\tif (read_ref(item->string, &oid_from)) {\n ++\t\t\t/*\n ++\t\t\t * The ref does not exist. The user probably\n ++\t\t\t * inserted a new 'update-ref' step with a new\n ++\t\t\t * branch name.\n ++\t\t\t */\n ++\t\t\toidcpy(&oid_from, the_hash_algo->null_oid);\n ++\t\t}\n ++\n ++\t\tres |= refs_update_ref(refs, \"rewritten during rebase\",\n ++\t\t\t\titem->string,\n ++\t\t\t\toid_to, &oid_from,\n +\t\t\t\t0, UPDATE_REFS_MSG_ON_ERR);\n ++\t}\n +\n -+\tfor (i = 0; i < ctx.nr; i++)\n -+\t\tfree(ctx.ref_names[i]);\n -+\tfree(ctx.ref_names);\n -+\tfree(ctx.old);\n -+\tfree(ctx.new);\n ++\tstring_list_clear(&refs_to_oids, 1);\n +\treturn res;\n +}\n +\n static int is_final_fixup(struct todo_list *todo_list)\n {\n \tint i = todo_list->current;\n +@@ sequencer.c: cleanup_head_ref:\n + \t\tstrbuf_release(&head_ref);\n + \t}\n + \n ++\tdo_update_refs(r);\n ++\n + \t/*\n + \t * Sequence of picks finished successfully; cleanup by\n + \t * removing the .git/sequencer directory\n +@@ sequencer.c: static int todo_list_add_update_ref_commands(struct todo_list *todo_list)\n + \t\t}\n + \t}\n + \n ++\twrite_update_refs_state(&ctx.refs_to_oids);\n ++\n + \tstring_list_clear(&ctx.refs_to_oids, 1);\n + \tfree(todo_list->items);\n + \ttodo_list->items = ctx.items;\n \n ## t/t3404-rebase-interactive.sh ##\n @@ t/t3404-rebase-interactive.sh: test_expect_success '--update-refs adds commands with --rebase-merges' '\n @@ t/t3404-rebase-interactive.sh: test_expect_success '--update-refs adds commands\n +\tgit branch -f third HEAD~1 &&\n +\ttest_commit extra2 fileX &&\n +\tgit commit --amend --fixup=L &&\n -+\t(\n -+\t\tgit rebase -i --autosquash --update-refs primary &&\n -+\n -+\t\tcompare_two_refs HEAD~3 refs/heads/first &&\n -+\t\tcompare_two_refs HEAD~3 refs/heads/second &&\n -+\t\tcompare_two_refs HEAD~1 refs/heads/third &&\n -+\t\tcompare_two_refs HEAD refs/heads/no-conflict-branch\n -+\t)\n ++\n ++\tgit rebase -i --autosquash --update-refs primary &&\n ++\n ++\tcompare_two_refs HEAD~3 refs/heads/first &&\n ++\tcompare_two_refs HEAD~3 refs/heads/second &&\n ++\tcompare_two_refs HEAD~1 refs/heads/third &&\n ++\tcompare_two_refs HEAD refs/heads/no-conflict-branch\n +'\n +\n # This must be the last test in this file\n 7: 3d7d3f656b4 ! 8: d2cfdbfc431 rebase: add rebase.updateRefs config option\n @@ Documentation/git-rebase.txt: start would be overridden by the presence of\n \tor point to a `squash! ...` or `fixup! ...` commit are not updated\n \tin this way.\n ++\n -+If the `--update-refs` option is enabled by default using the\n -+configuration variable `rebase.updateRefs`, this option can be\n -+used to override and disable this setting.\n ++If the configuration variable `rebase.updateRefs` is set, then this option\n ++can be used to override and disable this setting.\n \n INCOMPATIBLE OPTIONS\n --------------------\n\n-- \ngitgitgadget\n"},{"id":"458039","messageId":"72e0481b643e98c5670eee0bf5c199c4fb693b16.1656422759.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v3.git.1656422759.gitgitgadget@gmail.com","subject":"[PATCH v3 7/8] rebase: update refs from 'update-ref' commands","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-28T13:25:58Z","receivedAt":"2022-06-28T13:28:22Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe previous change introduced the 'git rebase --update-refs' option\nwhich added 'update-ref ' commands to the todo list of an\ninteractive rebase.\n\nTeach Git to record the HEAD position when reaching these 'update-ref'\ncommands. The ref/OID pair is stored in the\n$GIT_DIR/rebase-merge/update-refs file. A previous change parsed this\nfile to avoid having other processes updating the refs in that file\nwhile the rebase is in progress.\n\nNot only do we update the file when the sequencer reaches these\n'update-ref' commands, we then update the refs themselves at the end of\nthe rebase sequence. If the rebase is aborted before this final step,\nthen the refs are not updated.\n\nSigned-off-by: Derrick Stolee \n---\n sequencer.c | 114 +++++++++++++++++++++++++++++++++-\n t/t3404-rebase-interactive.sh | 23 +++++++\n 2 files changed, 136 insertions(+), 1 deletion(-)\n\ndiff --git a/sequencer.c b/sequencer.c\nindex 915d87a0336..4fd1c0b5bce 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -36,6 +36,7 @@\n #include \"rebase-interactive.h\"\n #include \"reset.h\"\n #include \"branch.h\"\n+#include \"log-tree.h\"\n \n #define GIT_REFLOG_ACTION \"GIT_REFLOG_ACTION\"\n \n@@ -148,6 +149,14 @@ static GIT_PATH_FUNC(rebase_path_squash_onto, \"rebase-merge/squash-onto\")\n */\n static GIT_PATH_FUNC(rebase_path_refs_to_delete, \"rebase-merge/refs-to-delete\")\n \n+/*\n+ * The update-refs file stores a list of refs that will be updated at the end\n+ * of the rebase sequence. The 'update-ref ' commands in the todo file\n+ * update the OIDs for the refs in this file, but the refs are not updated\n+ * until the end of the rebase sequence.\n+ */\n+static GIT_PATH_FUNC(rebase_path_update_refs, \"rebase-merge/update-refs\")\n+\n /*\n * The following files are written by git-rebase just after parsing the\n * command-line.\n@@ -4058,11 +4067,110 @@ leave_merge:\n \treturn ret;\n }\n \n-static int do_update_ref(struct repository *r, const char *ref_name)\n+static int write_update_refs_state(struct string_list *refs_to_oids)\n+{\n+\tint result = 0;\n+\tFILE *fp = NULL;\n+\tstruct string_list_item *item;\n+\tchar *path = xstrdup(rebase_path_update_refs());\n+\n+\tif (safe_create_leading_directories(path)) {\n+\t\tresult = error(_(\"unable to create leading directories of %s\"),\n+\t\t\t path);\n+\t\tgoto cleanup;\n+\t}\n+\n+\tfp = fopen(path, \"w\");\n+\tif (!fp) {\n+\t\tresult = error_errno(_(\"could not open '%s' for writing\"), path);\n+\t\tgoto cleanup;\n+\t}\n+\n+\tfor_each_string_list_item(item, refs_to_oids)\n+\t\tfprintf(fp, \"%s\\n%s\\n\", item->string, oid_to_hex(item->util));\n+\n+cleanup:\n+\tif (fp)\n+\t\tfclose(fp);\n+\treturn result;\n+}\n+\n+static int do_update_ref(struct repository *r, const char *refname)\n {\n+\tstruct string_list_item *item;\n+\tstruct string_list list = STRING_LIST_INIT_DUP;\n+\tint found = 0;\n+\n+\tsequencer_get_update_refs_state(r->gitdir, &list);\n+\n+\tfor_each_string_list_item(item, &list) {\n+\t\tif (!strcmp(item->string, refname)) {\n+\t\t\tstruct object_id oid;\n+\t\t\tfree(item->util);\n+\t\t\tfound = 1;\n+\n+\t\t\tif (!read_ref(\"HEAD\", &oid)) {\n+\t\t\t\titem->util = oiddup(&oid);\n+\t\t\t\tbreak;\n+\t\t\t}\n+\t\t}\n+\t}\n+\n+\tif (!found) {\n+\t\tstruct object_id oid;\n+\t\titem = string_list_append(&list, refname);\n+\n+\t\tif (!read_ref(\"HEAD\", &oid))\n+\t\t\titem->util = oiddup(&oid);\n+\t\telse\n+\t\t\titem->util = oiddup(the_hash_algo->null_oid);\n+\t}\n+\n+\twrite_update_refs_state(&list);\n+\tstring_list_clear(&list, 1);\n \treturn 0;\n }\n \n+static int do_update_refs(struct repository *r)\n+{\n+\tint res = 0;\n+\tstruct string_list_item *item;\n+\tstruct string_list refs_to_oids = STRING_LIST_INIT_DUP;\n+\tstruct ref_store *refs = get_main_ref_store(r);\n+\n+\tsequencer_get_update_refs_state(r->gitdir, &refs_to_oids);\n+\n+\tfor_each_string_list_item(item, &refs_to_oids) {\n+\t\tstruct object_id *oid_to = item->util;\n+\t\tstruct object_id oid_from;\n+\n+\t\tif (oideq(oid_to, the_hash_algo->null_oid)) {\n+\t\t\t/*\n+\t\t\t * Ref was not updated. User may have deleted the\n+\t\t\t * 'update-ref' step.\n+\t\t\t */\n+\t\t\tcontinue;\n+\t\t}\n+\n+\t\tif (read_ref(item->string, &oid_from)) {\n+\t\t\t/*\n+\t\t\t * The ref does not exist. The user probably\n+\t\t\t * inserted a new 'update-ref' step with a new\n+\t\t\t * branch name.\n+\t\t\t */\n+\t\t\toidcpy(&oid_from, the_hash_algo->null_oid);\n+\t\t}\n+\n+\t\tres |= refs_update_ref(refs, \"rewritten during rebase\",\n+\t\t\t\titem->string,\n+\t\t\t\toid_to, &oid_from,\n+\t\t\t\t0, UPDATE_REFS_MSG_ON_ERR);\n+\t}\n+\n+\tstring_list_clear(&refs_to_oids, 1);\n+\treturn res;\n+}\n+\n static int is_final_fixup(struct todo_list *todo_list)\n {\n \tint i = todo_list->current;\n@@ -4580,6 +4688,8 @@ cleanup_head_ref:\n \t\tstrbuf_release(&head_ref);\n \t}\n \n+\tdo_update_refs(r);\n+\n \t/*\n \t * Sequence of picks finished successfully; cleanup by\n \t * removing the .git/sequencer directory\n@@ -5709,6 +5819,8 @@ static int todo_list_add_update_ref_commands(struct todo_list *todo_list)\n \t\t}\n \t}\n \n+\twrite_update_refs_state(&ctx.refs_to_oids);\n+\n \tstring_list_clear(&ctx.refs_to_oids, 1);\n \tfree(todo_list->items);\n \ttodo_list->items = ctx.items;\ndiff --git a/t/t3404-rebase-interactive.sh b/t/t3404-rebase-interactive.sh\nindex 3cd20733bc8..63a69bc073e 100755\n--- a/t/t3404-rebase-interactive.sh\n+++ b/t/t3404-rebase-interactive.sh\n@@ -1813,6 +1813,29 @@ test_expect_success '--update-refs adds commands with --rebase-merges' '\n \t)\n '\n \n+compare_two_refs () {\n+\tgit rev-parse $1 >expect &&\n+\tgit rev-parse $2 >actual &&\n+\ttest_cmp expect actual\n+}\n+\n+test_expect_success '--update-refs updates refs correctly' '\n+\tgit checkout -B update-refs no-conflict-branch &&\n+\tgit branch -f base HEAD~4 &&\n+\tgit branch -f first HEAD~3 &&\n+\tgit branch -f second HEAD~3 &&\n+\tgit branch -f third HEAD~1 &&\n+\ttest_commit extra2 fileX &&\n+\tgit commit --amend --fixup=L &&\n+\n+\tgit rebase -i --autosquash --update-refs primary &&\n+\n+\tcompare_two_refs HEAD~3 refs/heads/first &&\n+\tcompare_two_refs HEAD~3 refs/heads/second &&\n+\tcompare_two_refs HEAD~1 refs/heads/third &&\n+\tcompare_two_refs HEAD refs/heads/no-conflict-branch\n+'\n+\n # This must be the last test in this file\n test_expect_success '$EDITOR and friends are unchanged' '\n \ttest_editor_unchanged\n-- \ngitgitgadget\n\n"},{"id":"458040","messageId":"e95ad41d3551244ea98c43145a089ecfd2536e54.1656422759.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v3.git.1656422759.gitgitgadget@gmail.com","subject":"[PATCH v3 5/8] sequencer: add update-ref command","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-28T13:25:56Z","receivedAt":"2022-06-28T13:28:24Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nAdd the boilerplate for an \"update-ref\" command in the sequencer. This\nconnects to the current no-op do_update_ref() which will be filled in\nafter more connections are created.\n\nThe syntax in the todo list will be \"update-ref \" to signal\nthat we should store the current commit as the value for updating\n at the end of the rebase.\n\nSigned-off-by: Derrick Stolee \n---\n rebase-interactive.c | 3 +++\n sequencer.c | 14 +++++++++++++-\n sequencer.h | 1 +\n 3 files changed, 17 insertions(+), 1 deletion(-)\n\ndiff --git a/rebase-interactive.c b/rebase-interactive.c\nindex 22394224faa..1ff07647af3 100644\n--- a/rebase-interactive.c\n+++ b/rebase-interactive.c\n@@ -57,6 +57,9 @@ void append_todo_help(int command_count,\n \" create a merge commit using the original merge commit's\\n\"\n \" message (or the oneline, if no original merge commit was\\n\"\n \" specified); use -c to reword the commit message\\n\"\n+\"u, update-ref = track a placeholder for the to be updated\\n\"\n+\" to this position in the new commits. The is\\n\"\n+\" updated at the end of the rebase\\n\"\n \"\\n\"\n \"These lines can be re-ordered; they are executed from top to bottom.\\n\");\n \tunsigned edit_todo = !(shortrevisions && shortonto);\ndiff --git a/sequencer.c b/sequencer.c\nindex 2ffee79619c..0b61835d441 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -1696,6 +1696,7 @@ static struct {\n \t[TODO_LABEL] = { 'l', \"label\" },\n \t[TODO_RESET] = { 't', \"reset\" },\n \t[TODO_MERGE] = { 'm', \"merge\" },\n+\t[TODO_UPDATE_REF] = { 'u', \"update-ref\" },\n \t[TODO_NOOP] = { 0, \"noop\" },\n \t[TODO_DROP] = { 'd', \"drop\" },\n \t[TODO_COMMENT] = { 0, NULL },\n@@ -2457,7 +2458,7 @@ static int parse_insn_line(struct repository *r, struct todo_item *item,\n \t\t\t command_to_string(item->command));\n \n \tif (item->command == TODO_EXEC || item->command == TODO_LABEL ||\n-\t item->command == TODO_RESET) {\n+\t item->command == TODO_RESET || item->command == TODO_UPDATE_REF) {\n \t\titem->commit = NULL;\n \t\titem->arg_offset = bol - buf;\n \t\titem->arg_len = (int)(eol - bol);\n@@ -4056,6 +4057,11 @@ leave_merge:\n \treturn ret;\n }\n \n+static int do_update_ref(struct repository *r, const char *ref_name)\n+{\n+\treturn 0;\n+}\n+\n static int is_final_fixup(struct todo_list *todo_list)\n {\n \tint i = todo_list->current;\n@@ -4431,6 +4437,12 @@ static int pick_commits(struct repository *r,\n \t\t\t\treturn error_with_patch(r, item->commit,\n \t\t\t\t\t\t\targ, item->arg_len,\n \t\t\t\t\t\t\topts, res, 0);\n+\t\t} else if (item->command == TODO_UPDATE_REF) {\n+\t\t\tstruct strbuf ref = STRBUF_INIT;\n+\t\t\tstrbuf_add(&ref, arg, item->arg_len);\n+\t\t\tif ((res = do_update_ref(r, ref.buf)))\n+\t\t\t\treschedule = 1;\n+\t\t\tstrbuf_release(&ref);\n \t\t} else if (!is_noop(item->command))\n \t\t\treturn error(_(\"unknown command %d\"), item->command);\n \ndiff --git a/sequencer.h b/sequencer.h\nindex 3ae541bb145..2cf5c1b9a38 100644\n--- a/sequencer.h\n+++ b/sequencer.h\n@@ -95,6 +95,7 @@ enum todo_command {\n \tTODO_LABEL,\n \tTODO_RESET,\n \tTODO_MERGE,\n+\tTODO_UPDATE_REF,\n \t/* commands that do nothing but are counted for reporting progress */\n \tTODO_NOOP,\n \tTODO_DROP,\n-- \ngitgitgadget\n\n"},{"id":"458041","messageId":"6528a50343f0fc3c0b2134f1af29cc30fab4f81f.1656422759.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v3.git.1656422759.gitgitgadget@gmail.com","subject":"[PATCH v3 4/8] sequencer: define array with enum values","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-28T13:25:55Z","receivedAt":"2022-06-28T13:28:26Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe todo_command_info array defines which strings match with which\ntodo_command enum values. The array is defined in the same order as the\nenum values, but if one changed without the other, then we would have\nunexpected results.\n\nMake it easier to see changes to the enum and this array by using the\nenum values as the indices of the array.\n\nSigned-off-by: Derrick Stolee \n---\n sequencer.c | 28 ++++++++++++++--------------\n 1 file changed, 14 insertions(+), 14 deletions(-)\n\ndiff --git a/sequencer.c b/sequencer.c\nindex 1094e146b99..2ffee79619c 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -1685,20 +1685,20 @@ static struct {\n \tchar c;\n \tconst char *str;\n } todo_command_info[] = {\n-\t{ 'p', \"pick\" },\n-\t{ 0, \"revert\" },\n-\t{ 'e', \"edit\" },\n-\t{ 'r', \"reword\" },\n-\t{ 'f', \"fixup\" },\n-\t{ 's', \"squash\" },\n-\t{ 'x', \"exec\" },\n-\t{ 'b', \"break\" },\n-\t{ 'l', \"label\" },\n-\t{ 't', \"reset\" },\n-\t{ 'm', \"merge\" },\n-\t{ 0, \"noop\" },\n-\t{ 'd', \"drop\" },\n-\t{ 0, NULL }\n+\t[TODO_PICK] = { 'p', \"pick\" },\n+\t[TODO_REVERT] = { 0, \"revert\" },\n+\t[TODO_EDIT] = { 'e', \"edit\" },\n+\t[TODO_REWORD] = { 'r', \"reword\" },\n+\t[TODO_FIXUP] = { 'f', \"fixup\" },\n+\t[TODO_SQUASH] = { 's', \"squash\" },\n+\t[TODO_EXEC] = { 'x', \"exec\" },\n+\t[TODO_BREAK] = { 'b', \"break\" },\n+\t[TODO_LABEL] = { 'l', \"label\" },\n+\t[TODO_RESET] = { 't', \"reset\" },\n+\t[TODO_MERGE] = { 'm', \"merge\" },\n+\t[TODO_NOOP] = { 0, \"noop\" },\n+\t[TODO_DROP] = { 'd', \"drop\" },\n+\t[TODO_COMMENT] = { 0, NULL },\n };\n \n static const char *command_to_string(const enum todo_command command)\n-- \ngitgitgadget\n\n"},{"id":"458043","messageId":"2bc647b6fcd6032a1e97e67739ced31117bfbfce.1656422759.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v3.git.1656422759.gitgitgadget@gmail.com","subject":"[PATCH v3 2/8] branch: consider refs under 'update-refs'","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-28T13:25:53Z","receivedAt":"2022-06-28T13:28:28Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe branch_checked_out() helper helps commands like 'git branch' and\n'git fetch' from overwriting refs that are currently checked out in\nother worktrees.\n\nA future update to 'git rebase' will introduce a new '--update-refs'\noption which will update the local refs that point to commits that are\nbeing rebased. To avoid collisions as the rebase completes, we want to\nmake the future data store for these refs to be considered by\nbranch_checked_out().\n\nThe data store is a plaintext file inside the 'rebase-merge' directory\nfor that worktree. The file alternates refnames and OIDs. The OIDs will\nbe used to store the to-be-written values as the rebase progresses, but\ncan be ignored at the moment.\n\nCreate a new sequencer_get_update_refs_state() method that parses this\nfile and populates a struct string_list with the ref-OID pairs. We can\nthen use this list to add to the current_checked_out_branches strmap\nused by branch_checked_out().\n\nTo properly navigate to the rebase directory for a given worktree,\nextract the static strbuf_worktree_gitdir() method to a public API\nmethod.\n\nWe can test that this works without having Git write this file by\nartificially creating one in our test script.\n\nSigned-off-by: Derrick Stolee \n---\n branch.c | 13 ++++++++++++\n sequencer.c | 42 +++++++++++++++++++++++++++++++++++++++\n sequencer.h | 9 +++++++++\n t/t2407-worktree-heads.sh | 22 ++++++++++++++++----\n 4 files changed, 82 insertions(+), 4 deletions(-)\n\ndiff --git a/branch.c b/branch.c\nindex 526e8237673..f252c4eef6c 100644\n--- a/branch.c\n+++ b/branch.c\n@@ -365,6 +365,7 @@ static void prepare_checked_out_branches(void)\n \t\tchar *old;\n \t\tstruct wt_status_state state = { 0 };\n \t\tstruct worktree *wt = worktrees[i++];\n+\t\tstruct string_list update_refs = STRING_LIST_INIT_DUP;\n \n \t\tif (wt->is_bare)\n \t\t\tcontinue;\n@@ -400,6 +401,18 @@ static void prepare_checked_out_branches(void)\n \t\t\tstrbuf_release(&ref);\n \t\t}\n \t\twt_status_state_free_buffers(&state);\n+\n+\t\tif (!sequencer_get_update_refs_state(get_worktree_git_dir(wt),\n+\t\t\t\t\t\t &update_refs)) {\n+\t\t\tstruct string_list_item *item;\n+\t\t\tfor_each_string_list_item(item, &update_refs) {\n+\t\t\t\told = strmap_put(¤t_checked_out_branches,\n+\t\t\t\t\t\t item->string,\n+\t\t\t\t\t\t xstrdup(wt->path));\n+\t\t\t\tfree(old);\n+\t\t\t}\n+\t\t\tstring_list_clear(&update_refs, 1);\n+\t\t}\n \t}\n \n \tfree_worktrees(worktrees);\ndiff --git a/sequencer.c b/sequencer.c\nindex 8c3ed3532ac..1094e146b99 100644\n--- a/sequencer.c\n+++ b/sequencer.c\n@@ -5901,3 +5901,45 @@ int sequencer_determine_whence(struct repository *r, enum commit_whence *whence)\n \n \treturn 0;\n }\n+\n+int sequencer_get_update_refs_state(const char *wt_dir,\n+\t\t\t\t struct string_list *refs)\n+{\n+\tint result = 0;\n+\tstruct stat st;\n+\tFILE *fp = NULL;\n+\tstruct strbuf ref = STRBUF_INIT;\n+\tstruct strbuf hash = STRBUF_INIT;\n+\tchar *path = xstrfmt(\"%s/rebase-merge/update-refs\", wt_dir);\n+\n+\tif (stat(path, &st))\n+\t\tgoto cleanup;\n+\n+\tfp = fopen(path, \"r\");\n+\tif (!fp)\n+\t\tgoto cleanup;\n+\n+\twhile (strbuf_getline(&ref, fp) != EOF) {\n+\t\tstruct object_id oid;\n+\t\tstruct string_list_item *item;\n+\n+\t\tif (strbuf_getline(&hash, fp) == EOF ||\n+\t\t get_oid_hex(hash.buf, &oid)) {\n+\t\t\twarning(_(\"update-refs file at '%s' is invalid\"),\n+\t\t\t\t path);\n+\t\t\tresult = -1;\n+\t\t\tgoto cleanup;\n+\t\t}\n+\n+\t\titem = string_list_append(refs, ref.buf);\n+\t\titem->util = oiddup(&oid);\n+\t}\n+\n+cleanup:\n+\tif (fp)\n+\t\tfclose(fp);\n+\tfree(path);\n+\tstrbuf_release(&ref);\n+\tstrbuf_release(&hash);\n+\treturn result;\n+}\ndiff --git a/sequencer.h b/sequencer.h\nindex da64473636b..3ae541bb145 100644\n--- a/sequencer.h\n+++ b/sequencer.h\n@@ -232,4 +232,13 @@ void sequencer_post_commit_cleanup(struct repository *r, int verbose);\n int sequencer_get_last_command(struct repository* r,\n \t\t\t enum replay_action *action);\n int sequencer_determine_whence(struct repository *r, enum commit_whence *whence);\n+\n+/**\n+ * Append the set of ref-OID pairs that are currently stored for the 'git\n+ * rebase --update-refs' feature if such a rebase is currently happening.\n+ *\n+ * Localized to a worktree's git dir.\n+ */\n+int sequencer_get_update_refs_state(const char *wt_dir, struct string_list *refs);\n+\n #endif /* SEQUENCER_H */\ndiff --git a/t/t2407-worktree-heads.sh b/t/t2407-worktree-heads.sh\nindex 4f59bc21303..f1e9e172a0c 100755\n--- a/t/t2407-worktree-heads.sh\n+++ b/t/t2407-worktree-heads.sh\n@@ -7,8 +7,11 @@ TEST_PASSES_SANITIZE_LEAK=true\n \n test_expect_success 'setup' '\n \ttest_commit init &&\n-\tgit branch -f fake-1 &&\n-\tgit branch -f fake-2 &&\n+\n+\tfor i in 1 2 3 4\n+\tdo\n+\t\tgit branch -f fake-$i || return 1\n+\tdone &&\n \n \tfor i in 1 2 3 4\n \tdo\n@@ -73,8 +76,19 @@ test_expect_success 'refuse to overwrite: worktree in rebase (merge)' '\n \techo refs/heads/fake-1 >.git/worktrees/wt-3/rebase-merge/head-name &&\n \techo refs/heads/fake-2 >.git/worktrees/wt-3/rebase-merge/onto &&\n \n-\ttest_must_fail git branch -f fake-1 HEAD 2>err &&\n-\tgrep \"cannot force update the branch '\\''fake-1'\\'' checked out at.*wt-3\" err\n+\tcat >.git/worktrees/wt-3/rebase-merge/update-refs <<-EOF &&\n+\trefs/heads/fake-3\n+\t$(git rev-parse HEAD~1)\n+\trefs/heads/fake-4\n+\t$(git rev-parse HEAD)\n+\tEOF\n+\n+\tfor i in 1 3 4\n+\tdo\n+\t\ttest_must_fail git branch -f fake-$i HEAD 2>err &&\n+\t\tgrep \"cannot force update the branch '\\''fake-$i'\\'' checked out at.*wt-3\" err ||\n+\t\t\treturn 1\n+\tdone\n '\n \n test_expect_success !SANITIZE_LEAK 'refuse to fetch over ref: checked out' '\n-- \ngitgitgadget\n\n"},{"id":"458042","messageId":"669f4abd59e7dbb13281e85144d085180934fd15.1656422759.git.gitgitgadget@gmail.com","threadId":"57949","inReplyTo":"pull.1247.v3.git.1656422759.gitgitgadget@gmail.com","subject":"[PATCH v3 3/8] rebase-interactive: update 'merge' description","fromName":"Derrick Stolee via GitGitGadget","fromEmail":"gitgitgadget@gmail.com","sentAt":"2022-06-28T13:25:54Z","receivedAt":"2022-06-28T13:28:30Z","isPatch":true,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"From: Derrick Stolee \n\nThe 'merge' command description for the todo list documentation in an\ninteractive rebase has multiple lines. The lines other than the first\none start with dots ('.') while the similar multi-line documentation for\n'fixup' does not.\n\nThe 'merge' command was documented when interactive rebase was first\nported to C in 145e05ac44b (rebase -i: rewrite append_todo_help() in C,\n2018-08-10). These dots might have been carried over from the previous\nshell implementation.\n\nThe 'fixup' command was documented more recently in 9e3cebd97cb (rebase\n-i: add fixup [-C | -c] command, 2021-01-29).\n\nLooking at the output in an editor, my personal opinion is that the dots\nare unnecessary and noisy. Remove them now before adding more commands\nwith multi-line documentation.\n\nSigned-off-by: Derrick Stolee \n---\n rebase-interactive.c | 6 +++---\n 1 file changed, 3 insertions(+), 3 deletions(-)\n\ndiff --git a/rebase-interactive.c b/rebase-interactive.c\nindex 87649d0c016..22394224faa 100644\n--- a/rebase-interactive.c\n+++ b/rebase-interactive.c\n@@ -54,9 +54,9 @@ void append_todo_help(int command_count,\n \"l, label