hypothesis tests are now failing?

[ctb]

with this error:

...
iff_cover.command_runner.CommandError: fatal: ambiguous argument 'origin/latest...HEAD': unknown revision or path not in the working tree.

which seems very weird to just suddenly happen. I don't think we recently updated anything that would cause this to break, but hypothesis is not in fact version pinned in setup.cfg so maybe it's being caused by a new version?

[ctb]

https://pypi.org/project/hypothesis/6.46.2/ just got released. maybe that's causing problems.

[mr-eyes]

I can see from here that git might be involved too.

https://stackoverflow.com/questions/45096755/fatal-ambiguous-argument-origin-unknown-revision-or-path-not-in-the-working

[ctb]

yeah... but that shouldn't have been a thing that suddenly changed! 🤔

[mr-eyes]

I have just tested it locally and it worked just fine. Can you try restarting the workflow?

[mr-eyes]

I restarted the workflow, and the problem is still there.
Originated from this line

sourmash/tox.ini

Line 160 in d218f6b

diff-cover --compare-branch {env:DIFF_AGAINST:origin/latest} {toxworkdir}/coverage.xml

[ctb]

(it's broken on multiple PRs)

[luizirber]

I think it was #2018, probably checkout@3 doesn't pick latest and the actual PR branch. Note that the ASV check already has a similar config (because it needs both the PR and latest available)