asb wrote: This is a user-facing change that definitely needs to be acknowledged in the release notes (clang/docs/ReleaseNotes.rst)
I agree with you that it seems more intuitive that a -march=rv32imaf invocation should default to ilp32f just as -march=rv32imafd defaults to ilp32d. I slightly disagree that this patch aligns us to GCC, because as I understood Kito's comment the GCC behaviour is really depending on what default ABI the toolchain was configured with. For a toolchain that defaults to ilp32d, presumably -march=rv32imaf still ends up with ilp32? That said, with the GCC cross-compilation model being so different to clang I don't think we necessarily need to match precisely, so I'm not opposed to a more intuitive default. https://github.com/llvm/llvm-project/pull/73489 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
