rust/compiler/rustc_trait_selection/src
Matthias Krüger 0470728e94
Rollup merge of #132084 - compiler-errors:param-env-with-err, r=lcnr,estebank
Consider param-env candidates even if they have errors

I added this logic in https://github.com/rust-lang/rust/pull/106309, but frankly I don't know why -- the logic was a very large hammer. It seems like recent changes to error tainting has made that no longer necessary.

Ideally we'd rework the way we handle error reporting in all of candidate assembly to be a bit more responsible; we're just suppressing candidates all willy-nilly and it leads to mysterious *other* errors cropping up, like the one that #132082 originally wanted to fix.

**N.B.** This has the side-effect of turning a failed resolution like `where Missing: Sized` into a trivial where clause that matches all types, but also I don't think it really matters?

I'm putting this up as an alternative to #132082, since that PR doesn't address the case when one desugars the APIT into a regular type param.

r? lcnr vibeck
2024-10-24 10:35:40 +02:00
..
error_reporting Deeply normalize type trace in type error reporting 2024-10-24 02:48:28 +00:00
errors Reformat using the new identifier sorting from rustfmt 2024-09-22 19:11:29 -04:00
solve remove unused field 2024-10-22 08:30:09 +02:00
traits Consider param-env candidates even if they have errors 2024-10-24 01:48:44 +00:00
errors.rs Reformat using the new identifier sorting from rustfmt 2024-09-22 19:11:29 -04:00
infer.rs move defining_opaque_types out of Canonical 2024-10-17 10:22:52 +02:00
lib.rs Stabilize the map/value methods on ControlFlow 2024-09-25 19:00:17 -07:00
regions.rs Reformat using the new identifier sorting from rustfmt 2024-09-22 19:11:29 -04:00
solve.rs impossible obligations check fast path 2024-10-10 06:09:50 -04:00