Patrick Steinhardt [off-list ref] writes:

[snip]

quoted

diff --git a/t/t5510-fetch.sh b/t/t5510-fetch.sh
index 4b113d7c27..a1ca4e1ac7 100755
--- a/t/t5510-fetch.sh
+++ b/t/t5510-fetch.sh

@@ -1639,6 +1639,94 @@ test_expect_success "backfill tags when providing a refspec" '
 	test_cmp expect actual
 '

+test_expect_success REFFILES "FETCH_HEAD is updated even if ref updates fail" '
+	test_when_finished rm -rf base repo &&
+
+	git init base &&
+	(
+		cd base &&
+		test_commit "updated" &&
+
+		git update-ref refs/heads/foo @ &&
+		git update-ref refs/heads/branch @
+	) &&
+
+	git init --bare repo &&
+	(
+		cd repo &&
+		rm -f FETCH_HEAD &&
+		git remote add origin ../base &&
+		>refs/heads/foo.lock &&

Hm. Is this compatible with all supported systems? We typically write
this as:

    : >refs/heads/foo.lock

But I have to acknowledge that I only do this because some people that
are more knowledgeable than I am know that we need this.

Other than that I'm happy with the current state of this patch series.
If the above turns out to be a non-issue I think it should be ready for
'next'.

I didn't know about this. The CI didn't complain about this too.

A quick search through our repo shows

$ rg --stats '^\s*>[\w/.-]+' t/
...
969 matches
969 matched lines
285 files contained matches
...

$ rg --stats '^\s*:\s*>[\w/.-]+' t/
...
188 matches
188 matched lines
58 files contained matches
...

So seems like we use both, meaning, this should be okay?

Thanks!

Patrick

Thanks for the review.

Since the other comment was a nit. I will hold off on sending a new version :)