Am 01.03.21 um 18:45 schrieb Junio C Hamano: > René Scharfe. <l.s.r@xxxxxx> writes: > >> Subject: [PATCH] t4205: assert %(describe) test coverage >> >> Document that the test is covering both describable and >> undescribable commits. >> >> Suggested-by: Ævar Arnfjörð Bjarmason <avarab@xxxxxxxxx> >> Signed-off-by: René Scharfe <l.s.r@xxxxxx> >> --- >> t/t4205-log-pretty-formats.sh | 11 ++++++++++- >> 1 file changed, 10 insertions(+), 1 deletion(-) >> >> diff --git a/t/t4205-log-pretty-formats.sh b/t/t4205-log-pretty-formats.sh >> index b47a0bd9eb..cabdf7d57a 100755 >> --- a/t/t4205-log-pretty-formats.sh >> +++ b/t/t4205-log-pretty-formats.sh >> @@ -965,8 +965,17 @@ test_expect_success 'log --pretty=reference is colored appropriately' ' >> test_expect_success '%(describe) vs git describe' ' >> git log --format="%H" | while read hash >> do >> - echo "$hash $(git describe $hash)" >> + if desc=$(git describe $hash) >> + then >> + : >expect-contains-good >> + else >> + : >expect-contains-bad >> + fi && >> + echo "$hash $desc" >> done >expect && >> + test_path_exists expect-contains-good && >> + test_path_exists expect-contains-bad && > > Hmph, I am not sure why we want temporary files for this (and I > doubt this "documenting" adds that much value to the tests to begin > with), but OK. Will queue. Variables would suffice, but make debugging harder. test_path_exists will at least print a suggestive file name. Perhaps we should add a test_assert? The added checks guard against neutering the test accidentally e.g. by tagging the currently undescribable commit in the setup phase. That would be hard to detect without it. > >> git log --format="%H %(describe)" >actual 2>err && >> test_cmp expect actual && >> test_must_be_empty err >> -- >> 2.30.1
