bcardosolopes wrote: > To be clear, this is a matter of idealism. There's no doubt that I can write > code to handle both types, and all the information I want is available. A > program processing CIR probably won't have a problem with this. It's more of > an issue for a person reading the CIR in text form and trying to understand > it. I guess the most I could say is that the difference between them could > lead a developer to introduce a bug due to their own misunderstanding.
This is super fair IMO, if we can be more clear and more consistent among operations (even if they are arguably orthogonal) that means accesses can be more uniform (or decoupled into operations that make it easier to reason), which is a good thing. https://github.com/llvm/llvm-project/pull/142041 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
