diff options
| author | Michael Goulet <michael@errs.io> | 2025-03-23 14:59:36 -0400 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-03-23 14:59:36 -0400 |
| commit | 045a1c78aed3e80386fc6f6734de32127cbe8e6d (patch) | |
| tree | 851cc4ccdba403612f9f4f23b00318b17bc92999 /compiler/rustc_codegen_llvm/src/common.rs | |
| parent | db52a4a03088a1b094723fc4fcb2176fcd4c19f2 (diff) | |
| parent | 77a106e61fe8331e646212382cd2ee5d2bbcd149 (diff) | |
| download | rust-045a1c78aed3e80386fc6f6734de32127cbe8e6d.tar.gz rust-045a1c78aed3e80386fc6f6734de32127cbe8e6d.zip | |
Rollup merge of #138861 - compiler-errors:flags-tweaks, r=lcnr
Tweak type flags, fix missing flags from coroutine kind ty Firstly, make sure we visit the coroutine kind ty. Since this kind ty is either infer (before upvar computation), or `()` or `i8`/`i16`/`i32`, this isn't really that big of a deal, since other types in the coroutine will also be infer, so we're not misreporting `ty.has_infer()` or anything, but it's still wrong not to do this. Furthermore, remove `HAS_TY_COROUTINE`, since nobody used it, and also remove special casing for `STILL_FURTHER_SPECIALIZABLE` since it's likely not important anymore? I have a vague recollection that it was important for polymorphization(?), but no tests seem to rely on this behavior. r? lcnr or reassign
Diffstat (limited to 'compiler/rustc_codegen_llvm/src/common.rs')
0 files changed, 0 insertions, 0 deletions