On Wed, Jan 10, 2024 at 06:52:30PM +0000, Justin Tobler via GitGitGadget wrote: > From: Justin Tobler <jltobler@xxxxxxxxx> > > Some tests set up reference locks by directly creating the lockfile. > While this works for the files reference backend, reftable reference > locks operate differently and are incompatible with this approach. > Generalize reference locking by preparing a reference transaction. As with the first patch, I think we could use d/f conflicts to get the same effect. Perhaps something like this: diff --git a/t/t5541-http-push-smart.sh b/t/t5541-http-push-smart.sh index df758e187d..7eb0e887e1 100755 --- a/t/t5541-http-push-smart.sh +++ b/t/t5541-http-push-smart.sh @@ -233,7 +233,8 @@ test_expect_success 'push --atomic fails on server-side errors' ' up="$HTTPD_URL"/smart/atomic-branches.git && # break ref updates for other on the remote site - mkdir "$d/refs/heads/other.lock" && + git -C "$d" update-ref -d refs/heads/other && + git -C "$d" update-ref refs/heads/other/block-me HEAD && # add the new commit to other git branch -f other collateral && @@ -244,12 +245,8 @@ test_expect_success 'push --atomic fails on server-side errors' ' # the new branch should not have been created upstream test_must_fail git -C "$d" show-ref --verify refs/heads/atomic && - # upstream should still reflect atomic2, the last thing we pushed - # successfully - git rev-parse atomic2 >expected && - # ...to other. - git -C "$d" rev-parse refs/heads/other >actual && - test_cmp expected actual && + # upstream should not have updated, as it cannot be written at all. + test_must_fail git -C "$d" rev-parse --verify refs/heads/other && # the new branch should not have been created upstream test_must_fail git -C "$d" show-ref --verify refs/heads/atomic && I do think that the original was slightly more interesting (since we could check that "other" still existed but was not updated), but I think the main point of the test is that "atomic" was not pushed at all. -Peff
