rjmccall added a comment. That doesn't seem like a good diagnostic. The actual problem is that you can't copy-construct a W from a volatile W&, but the error is being reported at a lower level than overload resolution; and even if you accept that as given, it's still complaining about the qualifier mismatch as if that's more important that the fact that the types don't match up.
I think this is probably just not the right place to bail out of the loop. Also, does the bug not reproduce if the S constructor's parameter isn't meaninglessly volatile-qualified? It's possible that we're just not stripping that properly in some place. http://reviews.llvm.org/D10881 _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
