| Message ID | cover.1684181855.git.code@khaugsbakk.name (mailing list archive) |
|---|---|
| Headers | show |
| Series | tag: keep the message file in case ref transaction fails | expand |
Kristoffer Haugsbakk <code@khaugsbakk.name> writes: > + GIT_EDITOR=./fakeeditor git tag -a foo && > -+ ! test -e .git/TAG_EDITMSG > ++ ! test_path_exists .git/TAG_EDITMSG This is not quite right. test_path_exists is loud when its expectation that the path _exists_ is not met, i.e. test_path_exists () { test "$#" -ne 1 && BUG "1 param" if ! test -e "$1" then echo "Path $1 doesn't exist" false fi } But this test expects that .git/TAG_EDITMSG to be missing. When the test is run with "-v" to make the output from this 'echo' visible, we will keep getting the complaint when the test is happy, which is not quite what we want. What you want to use is test_path_is_missing, without "!". > ## Notes (series) ## > - I tried to maintain the proper formatting by using `clang-format` via Emacs on > - the affected lines. > + I duplicated this message (this isn’t obvious in the diff): > + > + fprintf(stderr, > + _("The tag message has been left in %s\n"), > + path); > + > + Should this be factored into a static function instead? When the third copy is made, we would definitely insist avoiding copies, but until then, I am indifferent. Others may have different opinions, though.
§ Introduction (v2) The following material is the same compared to v1 up to but not including “Changes from the previous round (v1)” (The CI link is also new.) Cheers. ― The ref transaction can fail after the message has been written using the editor. The ref transaction is attempted after the message file (`TAG_EDITMSG`) has been unlinked, so there is no backup tag message file to retry the command.[1] This is unfortunate if someone has written more than e.g. “v1.99.4” in the editor. (I don’t know if people write long tag messages in practice.) Hold on to the tag message file until after the ref transaction in order to preserve the backup. † 1: On commit 91428f078b (The eighteenth batch, 2023-05-10) § Reproduction script ``` cd /tmp dir=$(mktemp -d) cd $dir git init git commit --allow-empty -mInit git tag release/v1 # Fails git tag -a release ``` Error message: ``` fatal: cannot lock ref 'refs/tags/release': 'refs/tags/release/v1' exists; cannot create 'refs/tags/release' ``` Better error message and behavior: ``` The tag message has been left in .git/TAG_EDITMSG fatal: cannot lock ref 'refs/tags/release': 'refs/tags/release/v1' exists; cannot create 'refs/tags/release' ``` § Alternatives considered My first thought was to find a way to “dry run” the ref update before opening the editor (the edge case of the ref update command succeeding the first time but not the second *real* time seemed incredibly unlikely to happen by happenstance, so I saw no reason to consider that). However that seemed like it would involve more code and conditionals, and I don’t know if the dry-run mode is even supported. A benefit of this alternative approach would be to error out immediately instead of opening the editor. But trying to create a tag which collides with an existing “namespace” seems very unlikely to happen in practice.[2] Losing a file is much worse than being inconvenienced to retry the command, so I decided to just focus on the former problem. Most importantly though this approach was within my ability to implement. † 2: Just observe my “Reproduction script”: one tries to create `release` after someone else made `release/v1`. But what is just “release”? What follows (next version) that? But why am I arguing against my change… § CI https://github.com/LemmingAvalanche/git/actions/runs/4984100467 § Changes compared to the previous round • Document `TAG_EDITMSG` • Improve test structure • Combine/squash the regression test for the change and the actual change • In the previous round the regression test commit would fail the test suite since it’s a `test_expect_success` and it comes before the actual change § Patches (compared to previous round) 1. (new) Document `TAG_EDITMSG` • So that we match `COMMIT_EDITMSG` 2. (was [1/3]) Test successful tag creation • Tweak commit message subject • Also remove `.git/TAG_EDITMSG` in `test_when_finished` 3. (was [2–3/3]) The main change plus a regression test • Tweak commit message and reflow paragraphs • Not a change but add a question for the reviewers (see the “Note”) Kristoffer Haugsbakk (3): doc: tag: document `TAG_EDITMSG` t/t7004-tag: add regression test for successful tag creation tag: keep the message file in case ref transaction fails Documentation/git-tag.txt | 10 ++++++++++ builtin/tag.c | 24 +++++++++++++++--------- t/t7004-tag.sh | 19 +++++++++++++++++++ 3 files changed, 44 insertions(+), 9 deletions(-) Range-diff against v1: -: ---------- > 1: 0e0e592853 doc: tag: document `TAG_EDITMSG` 1: 87b709d856 ! 2: aabeb4568e t/t7004-tag: add regression test for existing behavior @@ Metadata Author: Kristoffer Haugsbakk <code@khaugsbakk.name> ## Commit message ## - t/t7004-tag: add regression test for existing behavior + t/t7004-tag: add regression test for successful tag creation The standard tag message file is unlinked if the tag is created. @@ t/t7004-tag.sh: test_expect_success 'Does --[no-]contains stop at commits? Yes!' + echo Message >.git/TAG_EDITMSG + EOF + GIT_EDITOR=./fakeeditor git tag -a foo && -+ ! test -e .git/TAG_EDITMSG ++ ! test_path_exists .git/TAG_EDITMSG +' + test_done 2: 1f24aa43f7 < -: ---------- t/t7004-tag: add failing tag message file test 3: 999af290af ! 3: e67b6416b7 tag: keep the message file in case ref transaction fails @@ Metadata ## Commit message ## tag: keep the message file in case ref transaction fails - The ref transaction can fail after the user has written their tag message. In - particular, if there exists a tag `foo/bar` and `git tag -a foo` is said then - the command will only fail once it tries to write `refs/tags/foo`, which is - after one has closed the editor. + The ref transaction can fail after the user has written their tag + message. In particular, if there exists a tag `foo/bar` and `git tag -a + foo` is said then the command will only fail once it tries to write + `refs/tags/foo`, which is after the file has been unlinked. - Hold on to the message file for a little longer so that it is not unlinked - before the fatal error occurs. + Hold on to the message file for a little longer so that it is not + unlinked before the fatal error occurs. + Cc: Jeff King <peff@peff.net> Signed-off-by: Kristoffer Haugsbakk <code@khaugsbakk.name> ## Notes (series) ## - I tried to maintain the proper formatting by using `clang-format` via Emacs on - the affected lines. + I duplicated this message (this isn’t obvious in the diff): + + fprintf(stderr, + _("The tag message has been left in %s\n"), + path); + + Should this be factored into a static function instead? + + § Changes from previous round + + Squash (combine) the update to `tag.c` with the test so that the fix and + the regression test is added in one step. + + This makes more sense than what I was going for since the test suite + would fail on patch 2/3 of the previous round. + + Link: https://lore.kernel.org/git/xmqq4joeaxgw.fsf@gitster.g/T/#u ## builtin/tag.c ## @@ builtin/tag.c: static const char message_advice_nested_tag[] = @@ builtin/tag.c: int cmd_tag(int argc, const char **argv, const char *prefix) ref_transaction_free(transaction); if (force && !is_null_oid(&prev) && !oideq(&prev, &object)) printf(_("Updated tag '%s' (was %s)\n"), tag, + + ## t/t7004-tag.sh ## +@@ t/t7004-tag.sh: test_expect_success 'If tag is created then tag message file is unlinked' ' + ! test_path_exists .git/TAG_EDITMSG + ' + ++test_expect_success 'If tag cannot be created then tag message file is not unlinked' ' ++ test_when_finished "git tag -d foo/bar && rm .git/TAG_EDITMSG" && ++ write_script fakeeditor <<-\EOF && ++ echo Message >.git/TAG_EDITMSG ++ EOF ++ git tag foo/bar && ++ test_must_fail env GIT_EDITOR=./fakeeditor git tag -a foo && ++ test_path_exists .git/TAG_EDITMSG ++' ++ + test_done