zturner added a comment. In https://reviews.llvm.org/D53662#1279761, @jingham wrote:
> That depends on the definition of "fully qualified name". If you can ensure > that it means "name of C++ class" - or other ODR enforcing type system, then > you could make that assumption. In C you are free to redefine types on a > per-function basis if you so desire; and sadly some interface generators > (including MIG the one for generating Mach message handlers) do just this...) > So you would need to see a way to restrict the inputs to this function. > That doesn't seem like it would be straightforward to me. After I've changed up the codebase there is only one usage of it that I'm not sure is used in an ODR-enforcing context, which is `ObjCLanguageRuntime::LookupInCompleteClassCache`. All other usages are C++ or Java specific. Anyway, I guess I'll do it that way for now and upload the patch. This already turned out to be more invasvive than I was hoping for, so if this doesn't work out or someone doesn't like the approach, maybe I can just go back to the original patch and submit that https://reviews.llvm.org/D53662 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
