hokein wrote: > The change seems to be correct in the way code is written. > > However, I wonder what's the connection between the crash and the function > being changed? Is there a way to fail gracefully if there are more mistakes > in this function for incorrect code (e.g. fail deduction instead of crashing)? > > Is this also a correctness issue? Is it possible to come with an example > where the correct code causes a crash? (I suspect there is, we should > probably add it in as well as this is more severe than crashes on incorrect > code).
This issue comes from a bug in the implementation where we missed to handle the template template parameter. The resulting crash is a manifestation of this problem. I think it also affect correct code, I tried to come up with such example (which triggers the CTAD on template template parameters), but no luck :( https://github.com/llvm/llvm-project/pull/85405 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
