[PATCH] D49410: [PDB] Parse UDT symbols and pointers to members

[Aleksandr Urakov via Phabricator via llvm-commits]

aleksandr.urakov added a subscriber: zturner.
aleksandr.urakov added a comment.

Yes, I have understood that, thank you. It may be not trivial in a case of multiple nested unnamed unions/structs, but it's solvable for clang-emitted PDBs. But I'm not sure about cl-emitted PDBs too.

Can we leave the things as in this commit (and allow the fields overlapping in `MicrosoftRecordLayoutBuilder`)? So fields of nested unnamed UDTs will be treated as outer structure's fields. It will allow us to support both `clang` and `cl` PDBs, but what are the drawbacks of such solution?


https://reviews.llvm.org/D49410