================
@@ -401,12 +399,28 @@ static CompilerContext GetContextEntry(DWARFDIE die) {
return ctx(CompilerContextKind::Typedef);
case DW_TAG_base_type:
return ctx(CompilerContextKind::Builtin);
+ case DW_TAG_class_type:
+ case DW_TAG_structure_type:
+ case DW_TAG_union_type: {
+ CompilerContextKind kind = die.Tag() == DW_TAG_union_type
+ ? CompilerContextKind::Union
+ : CompilerContextKind::ClassOrStruct;
+ llvm::StringRef name = die.GetName();
+ if (!complete_template_names || name.contains('<'))
----------------
labath wrote:
You were probably asking @dwblaikie as I wasn't involved in the original
implementation. Anyway, here's what I think.
If we were able to simplify all template names, then I can see how such an
attribute could be useful as a replacement for the `contains('<')` heuristic,
though I believe that is just a performance optimization (avoid searching
through children DIEs if we see the name is templated) and shouldn't produce
false positives. It could also be used for things like "if all CUs are using
the same kind of names then don't bother trying to search for the other name
kind", though I'm unsure what difference would it make in practice (especially
the search with the full name should be very fast as it will just get zero
hits).
As it stands now, I think it's usefulness would be very limited. About the only
thing I can think of is skipping the query with the simplified name if all CUs
report they are using full names. That may make sense if you're seeing slowdown
due to the additional simplified-name queries.
https://github.com/llvm/llvm-project/pull/123054
_______________________________________________
lldb-commits mailing list
lldb-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
