aaron.ballman added a comment. In D130058#3697949 <https://reviews.llvm.org/D130058#3697949>, @smeenai wrote:
> Given that we have a non-obvious (at least to me) issue in a widely used > third-party library, would we consider giving users some way to opt out of > this error, at least as a transition tool? Thank you for the feedback! I agree that providing information like which enumeration type the value is out of range for would be super helpful. (Changing the note order might be significantly more involved but I can see how it would be helpful here.) Because the error is happening in boost, I think we have two paths forward: 1) targeted workaround for just boost, 2) downgrade the error to a warning-defaults-to-error. Because this is the second project to have run into some pain, I'm leaning towards downgrading the diagnostic for a release. One thing I would recommend is: add information to the release notes to set expectations about the warning being upgraded (if we downgrade it but don't tell users we're going to upgrade it to an error Real Soon Now, users will come to rely on the crutch). CHANGES SINCE LAST ACTION https://reviews.llvm.org/D130058/new/ https://reviews.llvm.org/D130058 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
