erichkeane wrote: The more i think about it... the more I wonder: should the 'backing variable' even BE in CIR? It is an implementation detail of the reference, so it really isn't representative of the AST/language itself, and is a detail that is perhaps part of lowering?
Though, I wonder how difficult that would make construction (probably very), so I could see reason to keep it around. That said, I'm thoroughly convinced that we should just represent the destruction as a part of the global reference, and keep this as a do-nothing-backing-store-only. Then I think it makes the rest of the implementation pretty trivial (just skips everything in lowering, since it has nothing to do). https://github.com/llvm/llvm-project/pull/200227 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
