aaron.ballman added a comment. In D137583#3917735 <https://reviews.llvm.org/D137583#3917735>, @dblaikie wrote:
> In D137583#3917706 <https://reviews.llvm.org/D137583#3917706>, @aaron.ballman > wrote: > >>> ...we expect template params to be fully qualified when comparing them for >>> simple template names >> >> So lldb is not inspecting the AST, they're doing reflection (of a sort) on >> the pretty printed names? Or am I misunderstanding something? > > Not reflection as such - but building names for the user, but partly from the > AST - basically LLDB wants to be able to produce the same name that > CGDebugInfo produces - so, maybe it should produce it the same way as > CGDebugInfo, which isn't to use the pretty printer from scratch. Ah thank you for the clarification, that makes more sense to me. > @aeubanks would this work for lldb's use case? > https://github.com/llvm/llvm-project/blob/main/clang/lib/CodeGen/CGDebugInfo.cpp#L5229 > it'd be identical to the original debug info generation, and looks like it > doesn't require a printing policy change/feature. Sorry I didn't think of > that earlier. I guess since `Qualified` would be `false` for lldb's use case, > you could go down into the implementation and just call the unqualified side > directly: `NamedDecl::printName(OS, Policy);` should print it unqualified for > this name, but respect the qualified printing policy flag for any nested > names, parameters, etc. I agree that it seems pretty sensible for the debugger to use the same mechanisms as debug info in terms of what names to display for users. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D137583/new/ https://reviews.llvm.org/D137583 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
