Refine some magic methods introspection #5273
Conversation
Tpt
force-pushed
the
tpt/magic-method-introspection
branch
from
fa5b044 to
e42bca5
Compare
- __richcmp__ is mapped to le/lt/ge/gt/eq/ne - __concat__, __repeat__, __inplace_concat__ and __inplace_repeat__ are mapped to their Python equivalent
Tpt
force-pushed
the
tpt/magic-method-introspection
branch
from
e42bca5 to
cc78c1b
Compare
davidhewitt
left a comment
davidhewitt
left a comment
There was a problem hiding this comment.
Thanks, looks good to me with a possible follow up!
| class EqDefaultNe: | ||
| def __eq__(self, /, other: EqDefaultNe) -> bool: ... |
There was a problem hiding this comment.
The __ne__ method gets "defaulted" on this type, is it just object.__ne__ (i.e. due to inheritance)?
Or is there actually still a unique EqDefaultNe.__ne__ method? We might want to special case here if so.
There was a problem hiding this comment.
This is a bit tricky: PyO3 generates its own __ne__ default implementation that negates __eq__. This is exactly the same behavior as object.__ne__. Adding an explicit EqDefaultNe.__ne__ would be indeed nicer but Mypy already property type check != with the current generated stubs because of the object.__ne__ fake inheritence. Happy to fix it in a follow up alongside the possible concat vs add conflict
There was a problem hiding this comment.
Totally makes sense, we can refine in a follow-up 👍
- __richcmp__ is mapped to le/lt/ge/gt/eq/ne - __concat__, __repeat__, __inplace_concat__ and __inplace_repeat__ are mapped to their Python equivalent
2 participants
__richcmp__is mapped to le/lt/ge/gt/eq/ne__concat__,__repeat__,__inplace_concat__and__inplace_repeat__are mapped to their Python equivalent. Such merge might create some naming conflicts but fixing it seems to require a quite big refactoring of the current approach (it will require some deduplication...). It is left to a future work.