rapidsna wrote: Thanks @AaronBallman!
> I think AttributedType might be a reasonable way forward. That's how we model > nullability qualifiers, for example: The difference from nullability qualifiers is that they are actually encoded as `AttributeKind`, whereas `__counted_by` has to store more than `Kind`, i.e., an expression and some dependent declarations. And the reason why we created a new sugar type instead of using `AttributedType` is that `AttributedType` itself doesn't have an extended data other than `AttributeKind` as an integer value. And more data seems to be stored in `TypeLoc`. Related to this, I remember we talked about having `CountAttributedType` to inherit `AttributedType`. I didn't want this to be picked up by `getAs<AttributedType>` so that we won't have to conditionalize `CountAttributedType` where `AttributeType` is generally handled and there's no other type inheriting `AttributedType` so far. E.g., BTagAttributedType is a sugar type that doesn't inherit `AttributedType`. https://github.com/llvm/llvm-project/pull/78000 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
