Repository navigation
Couple new rule suggestions #440
Description
I had a few thoughts for some more linter checks I'd like to have. Would be happy to implement them here if you guys are interested.
1. return <value> or yield <?from> <?value> in __init__()
Example:
class Example:
def __init__():
return 1 # bad
yield # bad
yield 1 # bad
yield from () # bad
return # okAny returning of values, or yielding, in an __init__() would result in a runtime error, but not until you try to instantiate an object of the class. So I think this should be a reasonably safe on-by-default error.
2. assert <constant or fixed value>
Normal asserts:
assert condition
assert condition, "message"There's already a flake8 check for asserting on a tuple specifically - e.g. accidentally adding parens around (condition, message). And there's a bugbear check for assert False warning that it's disabled by -O. But forgetting the condition (assert "message") or asserting on any other literal value or list/tuple will either always pass or always fail - and is probably unintentional?
3. Inconsistent returns in function
def func(x) -> int | None:
if x:
return 0
# should return None hereThis one's more opinionated, but if a function returns something in some case, it should explicitly return something (None) in all other cases.
Activity
cooperlees commentedon Dec 18, 2023
Thanks for asking before spending the time on the PR and us potentially saying nope.
With this one, are you only checking that there is always explicit returns at the end of a function's body? You're never worrying about the type right as I feel mypy / other type checkers will get that.
I don't feel the need for any of these to be optional. They all seems sane to me. But would love another contributor or users views ...
JelleZijlstra commentedon Jan 2, 2024
Agree 1 and 2 are good.
3 is probably better left to a type checker; mypy already has a similar check. In general, it requires type information to know whether a function always returns. Some examples: