matthias-springer wrote: > One question I have here is why not just insert right before the operation > that is being replaced? What is the need for trying to insert after one of > the values? Wouldnt' that remove the need for all of this complexity? It > gives you a singular place to insert.
Actually, I think you are right. I was thinking of a case where two ops (in different locations) were replaced with the same 1:N replacement values. And I thought that we store one mapping pair in that case. But we actually store one per original value (and not per replacement value). I just noticed this as I was trying to write down a counter-example... Let me update the PR, I think this will simplify things quite a bit. https://github.com/llvm/llvm-project/pull/115816 _______________________________________________ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
