Comment 7 for bug 2026319

The autopkgtest regression for git-remote-hg fails due to the patches git has to fix CVE-2022-39253, and is documented in bug 1993586, "Cannot add submodule using file transport".

Andreas had already forseen this in 2022:
https://bugs.launchpad.net/ubuntu/+source/git/+bug/1993586/comments/6
"The DEP8 tests need to be updated/changed, because they are all failing in git 2.38 in lunar currently, and I suspect the next non-security SRU for git will also trigger those in stable."

https://autopkgtest.ubuntu.com/results/autopkgtest-jammy/jammy/amd64/g/git-remote-hg/20230809_223307_203b4@/log.gz

...
521s Initialized empty Git repository in /tmp/autopkgtest.ZQXd3r/build.nXc/src/test/trash directory.main-push/tmp/gitrepo/.git/
522s Cloning into '/tmp/autopkgtest.ZQXd3r/build.nXc/src/test/trash directory.main-push/tmp/gitrepo/sub'...
522s fatal: transport 'file' not allowed
522s fatal: clone of '/tmp/autopkgtest.ZQXd3r/build.nXc/src/test/trash directory.main-push/tmp/sub' into submodule path '/tmp/autopkgtest.ZQXd3r/build.nXc/src/test/trash directory.main-push/tmp/gitrepo/sub' failed
522s not ok 52 - push with submodule
...

The testcase in git-remote-hg would ideally need to be changed, but since it would be a more or less no change upload apart from the testcase, it would likely get stuck in block-proposed due to not actually changing anything and introducing additional regression risk. I'll ask around to see if its worthwhile fixing the testcase or not.