efriedma-quic wrote: I don't understand the scenario you're envisioning here. If, for example, `sizeof(long)` is different in the AST vs. CodeGen, everything will very likely explode. And I don't want to make an open-ended commitment to support modifying properties of the target based on whatever happens to work at the moment.
What properties of the target can you safely manipulate? Would it make sense to store those properties in a separate datastructure? https://github.com/llvm/llvm-project/pull/88977 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
