RustAnalayzer mistakenly gives error on a.max(b)

[Ten0]

It seems to beleive max in a.max(b) is about Iterator::max, whereas it's about Ord::max, and as a consequence gives the error: expected 0 arguments, found 1 (and also when hovering with the mouse, doc points to core::iter::traits::iterator::Iterator). Using std::cmp::max(a, b) makes the error disappear.

Note that the feature that writes inferred types in light gray mentions &{unknown} instead of &Something (despite the type I'm iterating on being explicitly Vec<Something>), I suspect that is linked.

[Veykril]

It would be helpful to see a more complete snippet. Are you using any local imports(as in use statements that are inside of functions or expressions)? Because those are currently being ignored by RA which could lead to this problem.

[Ten0]

Are you using any local imports(as in use statements that are inside of functions or expressions)?

I'm not using use statements inside the function, however the Something I'm iterating on is declared in the function, just above, as it's a local structure used only to deserialize the result of a specific database query made there. Could that be the same issue?

Because those are currently being ignored by RA which could lead to this problem.

Why is that? :o

[Ten0]

Why is that? :o

Nvm, I found the answer.

#1165 (comment)

Looks like this issue is the long-standing-much-duplicated #1165 / #1559.

[Ten0]

Hmmm that being said, it looks like we could still improve the method inference when the type is {unknown}, by at least picking functions that have the proper number arguments/the right return type.

[flodiebold]

I don't know about that, but we should avoid giving follow-up errors when we have unknown types.

[flodiebold]

Closing in favour of #6975.