git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[PATCH v2 1/3] partial-clone: set default filter with --partial

From
Derrick Stolee via GitGitGadget <gitgitgadget@gmail.com>
Date
Apr 13, 2020, 15:25 UTC
Message-ID
<6f340d9aadf71d394ad320ad162f1d140b632f2c.1586791560.git.gitgitgadget@gmail.com>
In-Reply-To
<pull.745.v2.git.git.1586791560.gitgitgadget@gmail.com>
From: Derrick Stolee <dstolee@microsoft.com>

Partial clone means a lot of different things, including filtering out all blobs, large blobs, or objects along a certain pathspec. The pathspec option has limited uses, in particular due to the performance challenges in serving with such filters. The blob size option can be helpful for repositories with a small number of large binaries, but otherwise it is difficult to find a meaningful split between "small" and "large" blobs.

When I think of or recommend partial clone, I specifically mention the case of filtering out all blobs, and downloading those blobs only as needed.

This case is extremely useful, since it takes the best part of shallow clone (a very small initial download) without any of the downsides of restricted history.

However, the command-line interface can be confusing:
	git clone --filter=blob:none <url>
Add a simpler "--partial" option that defaults to this case:
	git clone --partial <url>

This should make the feature more discoverable. However, there is a significant interest in the size-limited filters as that behaves very similarly to Git LFS. For those cases, the following is available:

	git clone --partial=<size> <url>

There are quite a few commands using OPT_PARSE_LIST_OBJECTS_FILTER, includeing clone, fetch, and pack-objects. Augment this macro to include the "--partial[=<size>]" mode for free.

Modify the first partial clone test that checks --filter=blob:none to also check --partial with the same expected conditions after clone. The diff is much simpler to see when ignoring whitespace, since several lines added a leading tab. This test is essentially copied to include the two ways to specify a blob size limit of one byte.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 list-objects-filter-options.c | 18 +++++++++++++++
 list-objects-filter-options.h |  8 ++++++-
 t/t5616-partial-clone.sh      | 42 ++++++++++++++++++++++++++---------
 3 files changed, 57 insertions(+), 11 deletions(-)
diff --git a/list-objects-filter-options.c b/list-objects-filter-options.c
index 256bcfbdfe6..a71716ef75e 100644
--- a/list-objects-filter-options.c
+++ b/list-objects-filter-options.c
@@ -270,6 +270,24 @@ int opt_parse_list_objects_filter(const struct option *opt,
 	return 0;
 }
 
+int opt_set_blob_none_filter(const struct option *opt,
+			     const char *arg, int unset)
+{
+	struct strbuf filter_arg = STRBUF_INIT;
+	struct list_objects_filter_options *filter_options = opt->value;
+	
+	if (unset || !arg || !strcmp(arg, "0")) {
+		parse_list_objects_filter(filter_options, "blob:none");
+		return 0;
+	}
+	
+	strbuf_addf(&filter_arg, "blob:limit=%s", arg);
+	parse_list_objects_filter(filter_options, filter_arg.buf);
+	strbuf_release(&filter_arg);
+
+	return 0;
+}
+
 const char *list_objects_filter_spec(struct list_objects_filter_options *filter)
 {
 	if (!filter->filter_spec.nr)
diff --git a/list-objects-filter-options.h b/list-objects-filter-options.h
index 2ffb39222c4..ac38ffcbe86 100644
--- a/list-objects-filter-options.h
+++ b/list-objects-filter-options.h
@@ -62,6 +62,7 @@ struct list_objects_filter_options {
 
 /* Normalized command line arguments */
 #define CL_ARG__FILTER "filter"
+#define CL_ARG__PARTIAL "partial"
 
 void list_objects_filter_die_if_populated(
 	struct list_objects_filter_options *filter_options);
@@ -80,11 +81,16 @@ void parse_list_objects_filter(
 
 int opt_parse_list_objects_filter(const struct option *opt,
 				  const char *arg, int unset);
+int opt_set_blob_none_filter(const struct option *opt,
+			     const char *arg, int unset);
 
 #define OPT_PARSE_LIST_OBJECTS_FILTER(fo) \
 	{ OPTION_CALLBACK, 0, CL_ARG__FILTER, fo, N_("args"), \
 	  N_("object filtering"), 0, \
-	  opt_parse_list_objects_filter }
+	  opt_parse_list_objects_filter }, \
+	{ OPTION_CALLBACK, 0, CL_ARG__PARTIAL, fo, N_("size"), \
+	  N_("partial clone with blob filter"), \
+	  PARSE_OPT_OPTARG | PARSE_OPT_NONEG , opt_set_blob_none_filter }
 
 /*
  * Translates abbreviated numbers in the filter's filter_spec into their
diff --git a/t/t5616-partial-clone.sh b/t/t5616-partial-clone.sh
index 77bb91e9769..c42cef61296 100755
--- a/t/t5616-partial-clone.sh
+++ b/t/t5616-partial-clone.sh
@@ -33,17 +33,39 @@ test_expect_success 'setup bare clone for server' '
 # confirm we are missing all of the known blobs.
 # confirm partial clone was registered in the local config.
 test_expect_success 'do partial clone 1' '
-	git clone --no-checkout --filter=blob:none "file://$(pwd)/srv.bare" pc1 &&
-
-	git -C pc1 rev-list --quiet --objects --missing=print HEAD >revs &&
-	awk -f print_1.awk revs |
-	sed "s/?//" |
-	sort >observed.oids &&
+	for option in "--filter=blob:none" "--partial"
+	do
+		rm -rf pc1 &&
+		git clone --no-checkout "$option" "file://$(pwd)/srv.bare" pc1 &&
+
+		git -C pc1 rev-list --quiet --objects --missing=print HEAD >revs &&
+		awk -f print_1.awk revs |
+		sed "s/?//" |
+		sort >observed.oids &&
+
+		test_cmp expect_1.oids observed.oids &&
+		test "$(git -C pc1 config --local core.repositoryformatversion)" = "1" &&
+		test "$(git -C pc1 config --local remote.origin.promisor)" = "true" &&
+		test "$(git -C pc1 config --local remote.origin.partialclonefilter)" = "blob:none"
+	done
+'
 
-	test_cmp expect_1.oids observed.oids &&
-	test "$(git -C pc1 config --local core.repositoryformatversion)" = "1" &&
-	test "$(git -C pc1 config --local remote.origin.promisor)" = "true" &&
-	test "$(git -C pc1 config --local remote.origin.partialclonefilter)" = "blob:none"
+test_expect_success 'do partial clone with size limit' '
+	for option in "--filter=blob:limit=1" "--partial=1"
+	do
+		rm -rf pc-limit &&
+		git clone --no-checkout "$option" "file://$(pwd)/srv.bare" pc-limit &&
+
+		git -C pc-limit rev-list --quiet --objects --missing=print HEAD >revs &&
+		awk -f print_1.awk revs |
+		sed "s/?//" |
+		sort >observed.oids &&
+
+		test_cmp expect_1.oids observed.oids &&
+		test "$(git -C pc-limit config --local core.repositoryformatversion)" = "1" &&
+		test "$(git -C pc-limit config --local remote.origin.promisor)" = "true" &&
+		test "$(git -C pc-limit config --local remote.origin.partialclonefilter)" = "blob:limit=1"
+	done
 '
 
 test_expect_success 'verify that .promisor file contains refs fetched' '

-- 
gitgitgadget
Previous: Teng Long via GitGitGadgetNext: Derrick Stolee via GitGitGadget
Message 6 of 24 in “clone: document partial clone section”
  1. clone: document partial clone sectionTeng Long via GitGitGadget, Apr 2, 2020
  2. Derrick StoleeApr 2, 2020
  3. Junio C HamanoApr 2, 2020
  4. Derrick StoleeApr 2, 2020
  5. 0/3 clone: document partial clone sectionTeng Long via GitGitGadget, Apr 13, 2020
  6. 1/3 partial-clone: set default filter with --partialDerrick Stolee via GitGitGadget, Apr 13, 2020
  7. 2/3 clone: document --partial and --filter optionsDerrick Stolee via GitGitGadget, Apr 13, 2020
  8. 3/3 clone: document partial clone sectionDyrone Teng via GitGitGadget, Apr 13, 2020
  9. Philippe BlainOct 27, 2020
  10. Junio C HamanoApr 13, 2020
  11. Derrick StoleeApr 14, 2020
  12. Junio C HamanoApr 14, 2020
  13. Derrick StoleeApr 14, 2020
  14. Derrick StoleeApr 14, 2020
  15. clone: document partial clone sectionTeng Long via GitGitGadget, Oct 27, 2020
  16. Philippe BlainOct 27, 2020
  17. Junio C HamanoOct 27, 2020
  18. clone: document partial clone sectionTeng Long via GitGitGadget, Feb 25, 2021
  19. Philippe BlainFeb 25, 2021
  20. clone: document partial clone sectionTeng Long via GitGitGadget, Mar 2, 2021
  21. Junio C HamanoMar 3, 2021
  22. Fix inconsistent signed-off-by abd author nameTeng Long, May 6, 2021
  23. 1/1 clone: document partial clone sectionTeng Long, May 6, 2021
  24. Bagas SanjayaMay 7, 2021

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.