duplicated errors for invalid derives

[durka]

This code attempts to derive PartialOrd in an invalid situation, and is rewarded with a barrage of duplicate errors:

#[derive(PartialEq)] struct Comparable;
#[derive(PartialEq, PartialOrd)] struct Nope(Comparable);

The error the trait core::cmp::PartialOrd is not implemented for the type Comparable only appears once. However, the error binary operation < cannot be applied to type Comparable (along with note an implementation of std::cmp::PartialOrd might be missing for Comparable... gee, really?) appears eight times, four for < and four for >.

I assume this is because an error is generated at several points within the (invisible) generated PartialOrd implementation. But since they are all assigned the same span, maybe we can deduplicate.

[oli-obk]

Would the following be possible?:

1. encounter T: Trait bound failure
2. report error
3. make T: Trait true for the further compilation

It doesn't address all cases, but it would significantly improve this one.

[durka]

Should this be a UI or compile-fail test? I think cfail is out of favor these days, but a UI test will have to include a bug: #45519 (comment)

[michaelwoerister]

It should be a UI test, I'd say and we should fix that bug :)