dblaikie added a comment. In D101684#2737868 <https://reviews.llvm.org/D101684#2737868>, @tlively wrote:
> In D101684#2737842 <https://reviews.llvm.org/D101684#2737842>, @penzn wrote: > >> I think there is another dimension to this aside from project composition - >> intrinsics have a tendency to "interact" with their surroundings, and it >> better to capture the IR rather than the end result. Even if we can verify >> that simple calls produce instructions we expect, this might not hold true >> is the arguments change, or the call is in a different context. IR >> definitely gives more opportunities to test things through. > > Yeah, the contract that specific instructions are generated really only holds > in trivial cases by design. I'm not sure how to best formalize that, though. I don't know that much about intrinsics, but happy to help with test suggestions - do you have any practical examples of this you could show & I can see if I've got any ideas of good ways to test them? Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D101684/new/ https://reviews.llvm.org/D101684 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
