mirror of
https://github.com/rust-lang/rust.git
synced 2026-04-28 11:17:26 +03:00
fc758577be
`FindParamInClause` handle edge-cases If normalization is ambiguous, we want to treat the where-bound as non-global. The affected code should pretty much always have other errors in that case. It does make reasoning about trait solver cycles and rerunning more confusing and just feels wrong ™️ I encountered this while looking into `ml-kem` (https://github.com/rust-lang/trait-system-refactor-initiative/issues/246). I don't know whether this matters in practice and don't care to look into it too deeply. We also want to properly handle concrete aliases which mention to something mentioning a generic parameter due to another where-bound. r? @BoxyUwU