steakhal wrote: > Overall, I'd say that it's futile to try to recognize zero-sized types with a > "canonical type equal to" check, so you should just check whether > `ElemSizeInChars` is zero and do something based on that. (Either an early > return, or you can say `ElemSizeInChars = 1` at that point if you think that > that's the logically correct solution.)
Yes :/ > `<bikeshedding>`This way you could also avoid the immediately invoked lambda > in `getPointeeType` which is really ugly in my opinion.`</bikeshedding>` Alright. Duplicated the call I wanted to avoid, but I agree it looks better now. > _I only noticed that this PR was already merged after posting the review. > There is no need to revert the commit -- it's better than nothing -- but I'd > be happy if you created a followup change that also handles the testcase that > I mentioned._ I'm not sure what you refer to. This PR is not approved, hence not merged. Please continue with the review. https://github.com/llvm/llvm-project/pull/97199 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
