Properly track ImplObligations

[estebank]

Instead of probing for all possible impls that could have caused an
ImplObligation, keep track of its DefId and obligation spans for
accurate error reporting.

Follow to #89580. Addresses #89418.

[estebank]

Only the last commit is relevant. CC @nagisa, who reviewed the previous PR.

[petrochenkov]

r? @nagisa

[nagisa]

I have a hard time finding myself with enough time to give this PR the attention it deserves. Here are a couple comments I wrote last week; but otherwise you may want to consider looking for somebody who is able to dedicate a couple uninterrupted hours to review the PR.

Otherwise I'll just continue looking for an opportunity, but I can't promise it'll come soon.

[nagisa]

This reads weird. “because it appears within the type” is typically something one would expect to see for types with fields, but NoClone hasn't any.

[estebank]

Be aware that this is output only possible with feature(auto_traits) enabled.

[nagisa]

I had to look at this error message for a little while to figure out what

because of the requirements of the implementation of Bar for _

is trying to say here. In particular the wording to me parses to a causation of the following sort:

• We would use method <&Foo<T> as Bar>::foo, but cannot because there are unsatisfied trait bounds…
• In particular T: Bar bound from impl<T: Default + Bar> Bar for Foo<T> {} is not satisfied…
• Which was not satisfied because impl<T: Default + Bar> Bar for Foo<T> {} contains this bound.

whereas what we really mean here is:

• We would use method <&Foo<T> as Bar>::foo, but cannot because there are unsatisfied trait bounds…
• In particular T: Bar bound from impl<T: Default + Bar> Bar for Foo<T> {} is not satisfied…
• Which was not satisfied because T does not implement Bar..

[estebank]

The "duplicated" output is because it points at both Default and Bar bounds as unsatisfied for Bar, while it never talks about T: !Bar beyond the first mention. Would this read better?

error[E0599]: the method `foo` exists for reference `&Foo<T>`, but its trait bounds were not satisfied
  --> $DIR/missing-trait-bounds-for-method-call.rs:14:14
   |
LL | struct Foo<T> {
   | ------------- doesn't satisfy `Foo<T>: Bar`
...
LL |         self.foo();
   |              ^^^ method cannot be called on `&Foo<T>` due to unsatisfied trait bounds
   |
note: trait bound `T: Bar` was not satisfied
  --> $DIR/missing-trait-bounds-for-method-call.rs:10:19
   |
LL | impl<T: Default + Bar> Bar for Foo<T> {}
   |                   ^^^  ---     ------
   |                   |
   |                   unsatisfied trait bound introduced here
note: trait bound `T: Default` was not satisfied
  --> $DIR/missing-trait-bounds-for-method-call.rs:10:9
   |
LL | impl<T: Default + Bar> Bar for Foo<T> {}
   |         ^^^^^^^        ---     ------
   |         |
   |         unsatisfied trait bound introduced here
help: consider restricting the type parameters to satisfy the trait bounds
   |
LL | struct Foo<T> where T: Bar, T: Default {
   |               ++++++++++++++++++++++++

[estebank]

@oli-obk would you have time to take a look at this?

[oli-obk]

r? @oli-obk

[bors]

☀️ Try build successful - checks-actions
Build commit: 548649c9868f5cb125e19b7b4b74884aa901a5ae (548649c9868f5cb125e19b7b4b74884aa901a5ae)

[rust-timer]

Queued 548649c9868f5cb125e19b7b4b74884aa901a5ae with parent 37b55c8, future comparison URL.

[rust-timer]

Finished benchmarking commit (548649c9868f5cb125e19b7b4b74884aa901a5ae): comparison url.

Summary: This benchmark run shows 25 relevant regressions 😿 to instruction counts.

• Arithmetic mean of relevant regressions: 1.7%
• Largest regression in instruction counts: 3.9% on full builds of projection-caching check

If you disagree with this performance assessment, please file an issue in rust-lang/rustc-perf.

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR led to changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please fix the regressions and do another perf run. If the next run shows neutral or positive results, the label will be automatically removed.

@bors rollup=never
@rustbot label: +S-waiting-on-review -S-waiting-on-perf +perf-regression

[oli-obk]

@bors r+

[bors]

📌 Commit 5fd3786 has been approved by oli-obk

[bors]

⌛ Testing commit 5fd3786 with merge d2df372...

[bors]

☀️ Test successful - checks-actions
Approved by: oli-obk
Pushing d2df372 to master...

[rust-timer]

Finished benchmarking commit (d2df372): comparison url.

Summary: This benchmark run shows 19 relevant regressions 😿 to instruction counts.

• Arithmetic mean of relevant regressions: 1.9%
• Largest regression in instruction counts: 3.8% on full builds of projection-caching check

If you disagree with this performance assessment, please file an issue in rust-lang/rustc-perf.

Next Steps: If you can justify the regressions found in this perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please open an issue or create a new PR that fixes the regressions, add a comment linking to the newly created issue or PR, and then add the perf-regression-triaged label to this PR.

@rustbot label: +perf-regression

[pnkfelix]

Visiting for weekly performance triage.

The hit here was already anticipated. I trust that @oli-obk and @estebank will do what they can to improve performance around obligation causes

@rustbot label: +perf-regression-triaged