ChuanqiXu9 wrote: > > I feel they are workarounds. The `noload_redecls()` new interface may be > > the cleanest solution. > > For the first approach, I can potentially agree, even though we are already > deserializing with the current code so it can hardly be worse than that. For > the other two, can you please elaborate why you consider them workarounds? As > I pointed out, they are patterns that are already used in `ASTReader.cpp` and > `ASTReaderDecl.cpp`, for precisely the reasons we are running into now.
Maybe we have understanding for workarounds, but that doesn't matter. `noload_redecls` is pretty clear than the patterns you mentioned. > Also my understanding is that `getMostRecentExistingDecl` is implementation > wise exactly what you have in mind for `noload_redecls()`, minus the iterator > interface that I personally don't want to spend effort on. Yeah, if you already have this, why do you think it is burden to implement `noload_redecls()`? You've already reached it! https://github.com/llvm/llvm-project/pull/133057 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
