I'm hitting a weird case with `pants check` and i'...
# general
w
I'm hitting a weird case with
pants check
and i'm confused by the behaviour. On our main branch we run
pants check ::
, which passed for the HEAD commit. I have a branch that is update to date with main, with some changes, where if I run
pants check --changed-since=main --changed-dependents=transitive
I get valid check failures - but not on files i've changed on my branch. I have changed requirements, so i'm not surprised that it could invalidate files i've not worked directly on, which do need checking, and could have errors. However, i'm confused by the behaviour on the main branch, where i'd expect
pants check ::
to run against all files. I've tried deleting the mypy cache, lmdb cache, and i still get the same behaviour. It's concerning that
pants check ::
doesn't appear to be running everything it should be on our main branch. Am I missing something obvious here?
b
(Just checking, your last paragraph says
pants run check ::
, is that a typo for
pants check ::
?)
pants check ::
should indeed run against all files, so that does seem surprising. Are the check failures potentially related to the changed requirements?
w
@broad-processor-92400 I did - and yeah, it turns out it was an obtuse requirement change 🤦🏻 Sorry for the false alarm. All working as expected 👍
We ran into some issues with check/lint not picking up changes as expected in CI early in our pants adoption (caused by us misunderstanding some —changed-since semantics) and it left us a little wary of detection logic in pants - unfairly now
b
All good. Yeah, —changed can be a bit subtle at times