Support type[a.X] with qualified class names
#14825
Conversation
dcreager
requested review from
carljm,
MichaReiser,
AlexWaygood and
sharkdp
as code owners
| # TODO: no diagnostic | ||
| # error: [unresolved-attribute] | ||
| def f() -> type[a.b.C]: | ||
| # TODO: no diagnostic | ||
| # error: [unresolved-attribute] | ||
| return a.b.C |
There was a problem hiding this comment.
I left this as a TODO because I think this is not specific to type[]. The unresolved-attribute error I get is:
Type
<module 'System("/src/a/b.py")'>has no attributeb
which suggests that our import logic is binding a as a.b instead of a. Am I interpreting that right?
There was a problem hiding this comment.
Yes, it looks like you are! I think this is a previously-not-known bug, unless @AlexWaygood was already aware of it. Want to file an issue (labeled red-knot)? Could also be an interesting one for you to look into, would get you some exposure to the module resolver, which I don't think any of the other tasks do.
There was a problem hiding this comment.
I think I already knew about it, and I think we already have a TODO about it... whether we have an issue is another question... it's late here, so I won't look into whether we do or don't this evening :-)
There was a problem hiding this comment.
Opened #14826, can close it as a dup if it turns out we have another one already open
|
| # TODO: no diagnostic | ||
| # error: [unresolved-attribute] | ||
| def f() -> type[a.b.C]: | ||
| # TODO: no diagnostic | ||
| # error: [unresolved-attribute] | ||
| return a.b.C |
There was a problem hiding this comment.
Yes, it looks like you are! I think this is a previously-not-known bug, unless @AlexWaygood was already aware of it. Want to file an issue (labeled red-knot)? Could also be an interesting one for you to look into, would get you some exposure to the module resolver, which I don't think any of the other tasks do.
| let name_ty = self.infer_expression(slice); | ||
| if let Some(ClassLiteralType { class }) = name_ty.into_class_literal() { | ||
| Type::subclass_of(class) | ||
| } else { | ||
| todo_type!() | ||
| todo_type!("unsupported type[X] special form") | ||
| } | ||
| } | ||
| // TODO: attributes, unions, subscripts, etc. |
There was a problem hiding this comment.
we can remove "attributes" from this TODO
This adds support for
type[a.X], where thetypespecial form is applied to a qualified name that resolves to a class literal. This works for both nested classes and classes imported from another module.Closes #14545