Suggest the right help message for as_ref

[chenyukang]

Fixes#100605

[rust-highfive]

r? @compiler-errors

(rust-highfive has picked a reviewer for you, use r? to override)

[chenyukang]

I thought it will broke some tests,
surprised that we don't have a unittest to cover this error message.

[compiler-errors]

Thanks

@bors r+ rollup

[bors]

📌 Commit 3de74f7 has been approved by compiler-errors

It is now in the queue for this repository.

[WaffleLapkin]

I don't think this fixes the problem fully. We have at least 3 other unary operators, namely *, ! and - that can return arbitrary values and for them the diagnostic is still broken. Maybe we should to do something more general with method suggestions?...

[chenyukang]

I don't think this fixes the problem fully. We have at least 3 other unary operators, namely *, ! and - that can return arbitrary values and for them the diagnostic is still broken. Maybe we should to do something more general with method suggestions?...

Yes, I didn't consider these cases, this need some extra work.
Maybe we can remove this hard coded part in the help message:

*none.as_ref()

5 | takes_option(*none); // Suggests `*none.as_ref()`
| ------------ ^^^^^^^^^^^^^^
| | |
| | expected enum `Option`, found `&Option<String>`
| | help: you can convert from `&Option<T>` to `Option<&T>` using `.as_ref()`: `*none.as_ref()`
| arguments to this function are incorrect
|
= note: expected enum `Option<&String>`
found reference `&Option<String>`

Ofcourse, the right and best diagnostic is we need when to add the '( )'

(*none).as_ref()