Rewrite TypeId computation to not miss anything and work cross-crate. #35267
Conversation
|
r? @nrc (rust_highfive has picked a reviewer for you, use r? to override) |
| @@ -525,6 +393,142 @@ impl<'a, 'gcx, 'tcx> TyCtxt<'a, 'gcx, 'tcx> { | |||
| } | |||
| } | |||
|
|
|||
| struct TypeIdHasher<'a, 'gcx: 'a+'tcx, 'tcx: 'a> { | |||
| tcx: TyCtxt<'a, 'gcx, 'tcx>, | |||
| svh: &'a Svh, | |||
There was a problem hiding this comment.
Probably we should just be hashing the crate-name/crate-discriminant pair, rather than employing the SVH.
Why is that scary? It seems... right to me. The Note that the |
|
@bors r+ |
|
📌 Commit 7bef809 has been approved by |
|
@bors r=nikomatsakis |
|
📌 Commit d42da7b has been approved by |
Rewrite TypeId computation to not miss anything and work cross-crate. Fixes #33703 and also some soundness holes in `Any` due to `TypeId` ignoring most lifetimes.
Fixes #33703 and also some soundness holes in
Anydue toTypeIdignoring most lifetimes.