feat: Diagnose missing assoc items in trait impls

[Veykril]

No description provided.

[Veykril]

@bors r+

[bors]

📌 Commit 723d799 has been approved by Veykril

It is now in the queue for this repository.

[bors]

⌛ Testing commit 723d799 with merge c1e65aa...

[bors]

☀️ Test successful - checks-actions
Approved by: Veykril
Pushing c1e65aa to master...

[Veykril]

Just realized I forgot to replace the unwrap here. @Young-Flash since you are looking into #15909 already, do you mind changing this to ? and make this map a filter_map?

[Young-Flash]

sure

[lnicola]

[jnicholls]

This feature seems like a regression to me, or at the very least is not representative of compiler behavior. An associated const can have a default value on the trait, allowing implementations to inherit the default value. Since updating to v0.3.1740 I'm getting some false positive errors due to this feature.

Example of error-free (rustc) code: https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=4a8dad8097690736315ab7e64f19d0c9

trait MyTrait {
    const VALUE: &'static str = "Hello World";
}

struct MyType1;
impl MyTrait for MyType1 {}

struct MyType2;
impl MyTrait for MyType2 {
    const VALUE: &'static str = "Goodbye World";
}

fn main() {
    println!("{}", MyType1::VALUE);
    println!("{}", MyType2::VALUE);
}

The above code results in a rust-analyzer error on the impl MyTrait for MyType1 for not providing a value for the associated const VALUE.

[lnicola]

@jnicholls can you use the pre-release version until next week?

[jnicholls]

@lnicola Thanks the the wicked fast reply! Yes I can use the pre-release version, no problem.

[jnicholls]

Apologies, I didn't notice #15911 up above. Good deal! Have a great rest of your week.