J-Camilleri added a reviewer: aaron.ballman. J-Camilleri added a comment. Firstly, If there are other reviewers which I should include instead please let me know, the developers that changed around this area did so a long time ago and I do not know who to add.
I have described the problem in the github issue (https://github.com/llvm/llvm-project/issues/59684) and as I said there I am uncertain about my solution. Given the issue and my limited knowledge of the code base, I am thinking about three possible solutions to this issue. 1. Have the user defined operator present in the `IdResolver` so that it is returned with the implicitly declared declarations. - I have seen cases where the declaration is kept in an `FETokenInfo` section but not added to the `IdResolver`, it gets removed at some point and leads to it not being present when `CppLookupName` is called 2. Have the `IdResolver` declarations ordered by scope, going from most nested to least nested. - This suggestion stems from how the ommition of `Outer::invokeAssign` is handled. - Seems expensive to maintain. I do not understand the role of the `IdResolver` well so I do not know if the suggestions above misuse it/ change it's intended role. 3. Do not end the search at the `IdResolver` if we are searching for an 'operator='. - I went for this solution because it is simplest to implement and has the least changes. - This causes the `LookupResult` to temporary contain duplicate declarations, they are cleaned later by a call to `LookupResult::resolveKind` but it still feels like this should not happen. I guess we can skip the loop alltogether to not have duplications but I did not check to see if this is viable. I guess my main issue with this is understanding the role of `IdResolver`, should the order to implicit declarations change what it returns, should it already contain the user defined operator declaration? Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D147782/new/ https://reviews.llvm.org/D147782 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
