if ...: is now treated as if True: and it therefore thought tons of the eval_files contained unreachable statements. Replaced it with an arbitrary function call.
fix pyright errors
ignored error about Statement being unhashable, which it technically should be, although I guess we're inheriting a __hash__ from NamedTuple and it currently works :tm:. Need to change a bunch of equality checks in the code to make it less illegal, so leaving that for now.
TRIO910 "non-empty async function must contain checkpoint" did not properly detect
async def foo(): ...
as an empty function, since it was matching against the indented body. This is now the black style, which is why I found it. So fixed that, and updated test file to have both.
There's only a few more instances of IndentedBlock in the project, and none seem problematic, but it's possible this could be a problem in other places.
I can split this across multiple PRs, but idk if it's warranted. Now that we moved and can set up pre-commit it should be more incremental going forward.
jakkdl
commented
4 months ago
if ...:is now treated asif True:and it therefore thought tons of theeval_filescontained unreachable statements. Replaced it with an arbitrary function call.Statementbeing unhashable, which it technically should be, although I guess we're inheriting a__hash__fromNamedTupleand it currently works :tm:. Need to change a bunch of equality checks in the code to make it less illegal, so leaving that for now.as an empty function, since it was matching against the indented body. This is now the black style, which is why I found it. So fixed that, and updated test file to have both. There's only a few more instances of
IndentedBlockin the project, and none seem problematic, but it's possible this could be a problem in other places.I can split this across multiple PRs, but idk if it's warranted. Now that we moved and can set up pre-commit it should be more incremental going forward.