AaronBallman wrote: > > And `final` as well as `override`? (This is why I'm not convinced we should > > be backporting anything -- the problem is with printing in general and will > > crop up in various places, so we're not really fixing a regression so much > > as playing whack-a-mole with a few cases.) > > Why not? `override` and perhaps `final` output is broken so I'd say we fix > this on LLVM-18. It is better than leave it broken. I don't think this will > break many test cases once this bug got undetected.
"broken" is a bit subjective in this case -- this isn't a user-facing feature and it's always been best-effort. I realize a few folks would like to elevate it to be something more than best-effort, but it's still unclear how that works in practice and whether it's worth the ongoing maintenance burdens in upstream. Fixing a few corner cases does improve things and the fix is trivial to verify for correctness, so that's why I'm not strongly opposed. I'm just not certain it meets the usual criteria for inclusion in a dot release. https://github.com/llvm/llvm-project/pull/87281 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
