[PATCH v3 0/7] patch-id fixes and improvements
- From
Jerry Zhang via GitGitGadget <gitgitgadget@gmail.com>
- Date
- Oct 14, 2022, 08:56 UTC
- Message-ID
- <pull.1359.v3.git.1665737804.gitgitgadget@gmail.com>
- In-Reply-To
- <pull.1359.v2.git.1663654859.gitgitgadget@gmail.com>
These patches add fixes and features to the "git patch-id" command, mostly discovered through our usage of patch-id in the revup project (https://github.com/Skydio/revup). On top of that I've tried to make general cleanup changes where I can.
Summary:
1: Fixed a bug in the combination of --stable with binary files and header-only, and expanded the test to cover both binary and non-binary files.
2: Switch internal usage of patch-id in rebase / cherry-pick to use the stable variant to reduce the number of code paths and improve testing for bugs like above.
3: Fixed bugs with patch-id and binary diffs. Previously patch-id did not behave correctly for binary diffs regardless of whether "--binary" was given to "diff".
4: Fixed bugs with patch-id and mode changes. Previously mode changes were incorrectly excluded from the patch-id.
5: Add a new "--include-whitespace" mode to patch-id that prevents whitespace from being stripped during id calculation. Also add a config option for the same behavior.
6: Remove unused prefix from patch-id logic.
7: Update format-patch doc to specify when patch-ids are going to be equal to those generated by "git patch-id".
V1->V2: Fixed comment style V2->V3: The ---/+++ lines no longer get added to the patch-id of binary diffs. Also added patches 3-7 in the series.
Signed-off-by: Jerry Zhang jerry@skydio.com
Jerry Zhang (7):
patch-id: fix stable patch id for binary / header-only
patch-id: use stable patch-id for rebases
builtin: patch-id: fix patch-id with binary diffs
patch-id: fix patch-id for mode changes
builtin: patch-id: add --include-whitespace as a command mode
builtin: patch-id: remove unused diff-tree prefix
documentation: format-patch: clarify requirements for patch-ids to
matchDocumentation/git-format-patch.txt | 4 +- Documentation/git-patch-id.txt | 25 +++++-- builtin/log.c | 2 +- builtin/patch-id.c | 114 +++++++++++++++++++++-------- diff.c | 75 +++++++++---------- diff.h | 2 +- patch-ids.c | 10 +-- patch-ids.h | 2 +- t/t3419-rebase-patch-id.sh | 63 +++++++++++++--- t/t4204-patch-id.sh | 95 ++++++++++++++++++++++-- 10 files changed, 291 insertions(+), 101 deletions(-)
base-commit: d420dda0576340909c3faff364cfbd1485f70376 Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1359%2Fjerry-skydio%2Fjerry%2Frevup%2Fmaster%2Fpatch_ids-v3 Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1359/jerry-skydio/jerry/revup/master/patch_ids-v3 Pull-Request: https://github.com/gitgitgadget/git/pull/1359
Range-diff vs v2:
1: 945508df7b6 ! 1: 7d4c2e91ce0 patch-id: fix stable patch id for binary / header-only
@@ Commit message
populates the object ids, and normal which populates the text
diff. All branches will end up flushing the hunk.
+ Don't populate the ---a/ and +++b/ lines for binary diffs, to correspond
+ to those lines not being present in the "git diff" text output.
+ This is necessary because we advertise that the patch-id calculated
+ internally and used in format-patch is the same that what the
+ builtin "git patch-id" would produce when piped from a diff.
+
Update the test to run on both binary and normal files.
Signed-off-by: Jerry Zhang <jerry@skydio.com>
## diff.c ##
@@ diff.c: static int diff_get_patch_id(struct diff_options *options, struct object_id *oid
- the_hash_algo->update_fn(&ctx, p->two->path, len2);
+ if (p->one->mode == 0) {
+ patch_id_add_string(&ctx, "newfilemode");
+ patch_id_add_mode(&ctx, p->two->mode);
+- patch_id_add_string(&ctx, "---/dev/null");
+- patch_id_add_string(&ctx, "+++b/");
+- the_hash_algo->update_fn(&ctx, p->two->path, len2);
+ } else if (p->two->mode == 0) {
+ patch_id_add_string(&ctx, "deletedfilemode");
+ patch_id_add_mode(&ctx, p->one->mode);
+- patch_id_add_string(&ctx, "---a/");
+- the_hash_algo->update_fn(&ctx, p->one->path, len1);
+- patch_id_add_string(&ctx, "+++/dev/null");
+- } else {
+- patch_id_add_string(&ctx, "---a/");
+- the_hash_algo->update_fn(&ctx, p->one->path, len1);
+- patch_id_add_string(&ctx, "+++b/");
+- the_hash_algo->update_fn(&ctx, p->two->path, len2);
}
- if (diff_header_only)
@@ diff.c: static int diff_get_patch_id(struct diff_options *options, struct object
the_hash_algo->update_fn(&ctx, oid_to_hex(&p->two->oid),
the_hash_algo->hexsz);
- continue;
+- }
+-
+- xpp.flags = 0;
+- xecfg.ctxlen = 3;
+- xecfg.flags = XDL_EMIT_NO_HUNK_HDR;
+- if (xdi_diff_outf(&mf1, &mf2, NULL,
+- patch_id_consume, &data, &xpp, &xecfg))
+- return error("unable to generate patch-id diff for %s",
+- p->one->path);
+ } else {
++ if (p->one->mode == 0) {
++ patch_id_add_string(&ctx, "---/dev/null");
++ patch_id_add_string(&ctx, "+++b/");
++ the_hash_algo->update_fn(&ctx, p->two->path, len2);
++ } else if (p->two->mode == 0) {
++ patch_id_add_string(&ctx, "---a/");
++ the_hash_algo->update_fn(&ctx, p->one->path, len1);
++ patch_id_add_string(&ctx, "+++/dev/null");
++ } else {
++ patch_id_add_string(&ctx, "---a/");
++ the_hash_algo->update_fn(&ctx, p->one->path, len1);
++ patch_id_add_string(&ctx, "+++b/");
++ the_hash_algo->update_fn(&ctx, p->two->path, len2);
++ }
+
+ if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
+ fill_mmfile(options->repo, &mf2, p->two) < 0)
+ return error("unable to read files to diff");
@@ diff.c: static int diff_get_patch_id(struct diff_options *options, struct object
+ patch_id_consume, &data, &xpp, &xecfg))
+ return error("unable to generate patch-id diff for %s",
+ p->one->path);
- }
--
-- xpp.flags = 0;
-- xecfg.ctxlen = 3;
-- xecfg.flags = XDL_EMIT_NO_HUNK_HDR;
-- if (xdi_diff_outf(&mf1, &mf2, NULL,
-- patch_id_consume, &data, &xpp, &xecfg))
-- return error("unable to generate patch-id diff for %s",
-- p->one->path);
--
++ }
if (stable)
flush_one_hunk(oid, &ctx);
}
## t/t3419-rebase-patch-id.sh ##
@@ t/t3419-rebase-patch-id.sh: test_expect_success 'setup: 500 lines' '
- git cherry-pick main >/dev/null 2>&1
- '
+ git add newfile &&
+ git commit -q -m "add small file" &&
+
+- git cherry-pick main >/dev/null 2>&1
+-'
++ git cherry-pick main >/dev/null 2>&1 &&
-test_expect_success 'setup attributes' '
- echo "file binary" >.gitattributes
--'
--
++ git branch -f squashed main &&
++ git checkout -q -f squashed &&
++ git reset -q --soft HEAD~2 &&
++ git commit -q -m squashed
+ '
+
test_expect_success 'detect upstream patch' '
- git checkout -q main &&
+- git checkout -q main &&
++ git checkout -q main^{} &&
scramble file &&
+ git add file &&
+ git commit -q -m "change big file again" &&
@@ t/t3419-rebase-patch-id.sh: test_expect_success 'detect upstream patch' '
- git checkout -q other^{} &&
- git rebase main &&
- git rev-list main...HEAD~ >revs &&
-- test_must_be_empty revs
-+ test_must_be_empty revs &&
+ test_must_be_empty revs
+ '
+
++test_expect_success 'detect upstream patch binary' '
+ echo "file binary" >.gitattributes &&
+ git checkout -q other^{} &&
+ git rebase main &&
+ git rev-list main...HEAD~ >revs &&
+ test_must_be_empty revs &&
-+ rm .gitattributes
- '
-
++ test_when_finished "rm .gitattributes"
++'
++
test_expect_success 'do not drop patch' '
-@@ t/t3419-rebase-patch-id.sh: test_expect_success 'do not drop patch' '
- git commit -q -m squashed &&
+- git branch -f squashed main &&
+- git checkout -q -f squashed &&
+- git reset -q --soft HEAD~2 &&
+- git commit -q -m squashed &&
git checkout -q other^{} &&
test_must_fail git rebase squashed &&
- git rebase --quit
-+ git rebase --abort &&
++ test_when_finished "git rebase --abort"
++'
++
++test_expect_success 'do not drop patch binary' '
+ echo "file binary" >.gitattributes &&
+ git checkout -q other^{} &&
+ test_must_fail git rebase squashed &&
-+ git rebase --abort &&
-+ rm .gitattributes
++ test_when_finished "git rebase --abort" &&
++ test_when_finished "rm .gitattributes"
'
test_done
2: 30ec43cd129 ! 2: 25e28b7dab3 patch-id: use stable patch-id for rebases
@@ Commit message
patch-id: use stable patch-id for rebases
Git doesn't persist patch-ids during the rebase process, so there is
- no need to specifically invoke the unstable variant.
-
- This allows the legacy unstable id logic to be cleaned up.
+ no need to specifically invoke the unstable variant. Use the stable
+ logic for all internal patch-id calculations to minimize the number of
+ code paths and improve test coverage.
Signed-off-by: Jerry Zhang <jerry@skydio.com>
-: ----------- > 3: 21642128927 builtin: patch-id: fix patch-id with binary diffs
-: ----------- > 4: 6e07cfd5691 patch-id: fix patch-id for mode changes
-: ----------- > 5: bbaa2425ad0 builtin: patch-id: add --include-whitespace as a command mode
-: ----------- > 6: a1f6f36d487 builtin: patch-id: remove unused diff-tree prefix
-: ----------- > 7: 69440797f30 documentation: format-patch: clarify requirements for patch-ids to match-- gitgitgadget