Michael137 wrote: > I think the problem here is of a more fundamental nature. `FindTypes` finds > **types**. `bar` is _not_ a type. `bar<int>` is. So, while a type query for > `bar` will find the definition DIE with `DW_AT_name="bar"`, that DIE will > actually be defining the type `bar<int>`. So when the implementation looks at > the type constructed from that DIE, it will see that the name does not match > what is being asked (it gets this name through a different API, so it will > include the template args even without this patch), and discard the type. > > As far as i know, this is all WAI. Existing FindType callers expect to get a > specific type as a result of their query -- not a collection of > instantiations of that template. (That's definitely true in this case, where > returning the collection of instantiations would just push the burden of > filtering them onto the caller.) That's not to say this kind of a template > search is not useful. Having something like that would go a long way towards > making expressions like `(bar<int> *) ptr` work. But that would probably be a > different API and the problem is that this is basically impossible to > implement in a non-simplified-template-name world, and even with simplified > names, returning all instantiations might be too expensive.
Makes sense, thanks for elaborating LGTM The `TestDAP_evaluate.py` failure is https://github.com/llvm/llvm-project/issues/116041 (which should be fixed once you rebase) https://github.com/llvm/llvm-project/pull/116068 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
