AaronBallman wrote: > My feeling is that it's not worth the complexity. Getting an accurate > benchmak would be difficult, it would only manifest in extremely templated > code I wonder if @AaronBallman has opinion.
I think the complexity may be worth it -- FWIW, this originally came from https://github.com/llvm/llvm-project/commit/6163aa96799cbad7f2f58e02c5bebee9647056a5. CC @zygoloid for opinions as the original author. One question I have is whether it would make sense to modify `DestroyTemplateIdAnnotationsRAIIObj` to not call `MaybeDestroyTemplateIds()` in the first place, rather than add a new RAII object that early returns from `MaybeDestroyTemplateIds()`> so that the logic is more self-contained? https://github.com/llvm/llvm-project/pull/89494 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
