AaronBallman wrote: > I don't think the discussion here has run its course > https://github.com/llvm/llvm-project/pull/115416
Yeah, I had explicitly asked for a review from @Bigcheese before that landed, so I was surprised to see that get merged. > I don't think we should force users for our too conservative decisions. As I > said there are many false positive error messages which doesn't affect the > process actually. False positive *error* diagnostics? Warnings can have false positives; errors are not allowed to and that suggests we have some serious issues elsewhere to address first. > I feel it is really wishful thinking to ban this and hope we can be in a more > safe world. I think it may only stop more people using modules and change > nothing else. Again, please give users a chance to choose what they like. Maybe I'm misunderstanding something but that sounds like the exact opposite of how we usually operate. We typically do aim for improved safety and if that means people don't use modules as much, that's fine -- it's better for us to be too restrictive and then relax those restrictions in the future than to be too permissive and have to support that forever due to users relying on it. https://github.com/llvm/llvm-project/pull/117840 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
