Duplicate deref-chain method names

[A4-Tacks]

rust-analyzer version: rust-analyzer 0.4.2629-standalone

rustc version: rustc 1.91.0-nightly (040a98a 2025-08-20)

editor or extension: Vim

code snippet to reproduce:

fn foo(x: Vec<i32>) {
    x.is|
}

Output completion list:

is_empty()~ const fn(&self) -> bool       f
is_empty()~ const fn(&self) -> bool       f
is_sorted()~ fn(&self) -> bool            f
is_sorted_by(…)~ fn(&'a self, F) -> bool  f

[ChayimFriedman2]

They are not the same method, and generally we have no way to know that they do the same thing.

[A4-Tacks]

According to the method search order, methods with the same name can be directly excluded

[A4-Tacks]

Sorry, I filled in the r-a version incorrectly

There are no bugs in rust-analyzer 1.91.0-nightly (040a98a 2025-08-20), while rust-analyzer 0.4.2629-standalone has them

[flodiebold]

Yeah, I think if we provide both completions one of them should actually result in fully qualified method call syntax so that it actually refers to the 'shadowed' method.

[A4-Tacks]

Yeah, I think if we provide both completions one of them should actually result in fully qualified method call syntax so that it actually refers to the 'shadowed' method.

If it is possible to complete 'qualified method call', it is best to disable it by default in user settings,
as the 'shadowed' methods is rarely called but always show

ALSO, compared to 'qualified method call', explicit dereference calls may be better, considering opaque-types

[ChayimFriedman2]

Sorry, I filled in the r-a version incorrectly

There are no bugs in rust-analyzer 1.91.0-nightly (040a98a 2025-08-20), while rust-analyzer 0.4.2629-standalone has them

Yes, this is #20587.

I agree we should make the completion actually call the correct method, and also clearly show the difference in the completion list, but I don't think we should omit it.