Artem-B wrote: > I don't think we should rely on these on the host at all, the addition was a > design mistake initially, we probably should not double down on it.
I agree with it in principle. However, removing things that already exist should be done with consideration for the existing users. This particular macro may happen to be working well enough for some users and we may want to consider whether we can transition from here to the point where these macros no longer exist on the host, without causing much disruption, or, at least, giving the affected users enough of heads-up time to deal with it. That said, I do not have enough context on how widely those macros are used in the actual HIP code. I can see scenarios where just disabling the macro on the host side may be a sensible choice, but, if the macro was publicly documented, then it should probably go through the standard deprecation process -- document deprecation intent/add warnings in the next release, remove it in the release after that. It's mostly a logistical problem for HIP owners, not technical one. https://github.com/llvm/llvm-project/pull/109663 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
