Repository navigation
support human-friendly names in --disable, noqa and --enable #236
Description
Ruff might be adding support for human-friendly names for rules astral-sh/ruff#1773
So at some point somebody probably needs to come up with names for our rules as they're implemented in Ruff anyway, in which case we might as well do it ourselves*. The easiest place to support this would be in --disable and --enable. We don't currently parse # noqa, see #185, but this adds another reason to handle noqa-parsing on our own.
But the main upside is probably in daily speech and discussion, throwing around ASYNC102 or ASYNC119 ~always requires looking up the codes even for me.
* Claude did decent but not great when asked for suggestions:
ASYNC100:
- timeout-no-await
- idle-timeout
- checkpoint-missing
- await-required
ASYNC101:
- unsafe-nursery-yield
- yield-breaks-nursery
- yield-context-only
- invalid-yield
ASYNC102:
- unshielded-await
- await-needs-shield
- unsafe-finally-await
- unprotected-await
ASYNC103:
- unhandled-cancellation
- cancellation-not-reraised
- missing-cancel-handler
- reraise-required
ASYNC104:
- unreraised-cancellation
- must-reraise-cancel
- cancellation-not-propagated
- incomplete-cancellation
Activity
jakkdl commentedon May 24, 2024
--enableand--disablenoqa.* This would depend on full
# noqasupport #185, as we currently fully rely on flake8 for noqa suppression.* It would also introduce additional incompatibility with flake8-noqa, as they would flag invalid-looking noqa comments. (unless we suggest upstream to update their regex.. which might not be impossible with ruff considering human-friendly). Add support for external codes plinss/flake8-noqa#30 might resolve this, perhaps with a followup to enable non-code-looking rules, but it's been open for 6 months with no response from repo owner.
[-]Come up with human-friendly names for all error codes, and support those in `--disable` etc[/-][+]support human-friendly names in `--disable`, `noqa` and `--enable`[/+]