squawk: ignore require-statement-timeout and require-lock-timeout rules - #4920
Conversation
|
instead of this, should we just use a config file: |
|
Where do you think we should put it? Repo root? |
Yh repo root would be the ideal place or could even sit in the scripts dir |
c17037c to
8276155
Compare
|
Going to have to temporarily add a migration that would set these off.. |
|
Oh and this was only split into two rules in 2.60.0 (https://github.com/sbdchd/squawk/blob/master/CHANGELOG.md#v2600---2026-07-18), so I should probably upgrade us to that at the same time. |
8276155 to
c2ea754
Compare
| # This file was automatically copied from notifications-utils@121.0.0 | ||
|
|
||
| exclude-newer = "7 days" | ||
| exclude-newer = "5 days" |
There was a problem hiding this comment.
There was a problem hiding this comment.
Yup that is just a very temporary hack - in reality putting an exception here doesn't work very well because the copy_config() thing overwrites the uv.toml before it runs the compile for requirements_for_test.txt
There was a problem hiding this comment.
Which raises an interesting point actually - the compile steps for requirements.txt and requirements_for_test.txt run against different uv.toml versions.
There was a problem hiding this comment.
In theory what we should do is re-run the compile for requirements.txt after the copy_config() and repeat the process until nothing changes and we've therefore reached a "fixed point" (https://en.wikipedia.org/wiki/Fixed-point_combinator)
There was a problem hiding this comment.
I guess ideally the freeze command would:
- install specified version of utils
- run
copy_config() - freeze both
.txtfiles andsync
There was a problem hiding this comment.
Might work if you're certain utils is the only thing that affects the outcome.
f40b374 to
c2ea754
Compare
c2ea754 to
bad5d86
Compare
3 participants
These are not appropriate for us, because we already set these for each migration (though in a way squawk can't see - in the harness) and it prompts people to add these lines to migrations without really knowing what values to give them just to shut the CI up.