joey marked an inline comment as done. joey added inline comments.
================ Comment at: lib/Sema/SemaType.cpp:1224 } - return ""; + return OpenCLAccessAttr::Keyword_read_only; } ---------------- aaron.ballman wrote: > The caller can no longer tell the difference between a real-only OpenCL > access attribute and no OpenCL access attribute. I know that this was the > effective behavior of the original code, but that's specific to the current > use case, so I think this change is a tiny regression in the semantics. I renamed this function. ================ Comment at: lib/Sema/SemaType.cpp:1629 + Result = Context.Id##ROTy; break; \ + default: assert(0 && "Unknown access attribute!"); \ + } \ ---------------- aaron.ballman wrote: > Just checking, but, has this situation already been diagnosed elsewhere? Maybe I should actually replace this with an llvm_unreachable? Repository: rL LLVM https://reviews.llvm.org/D26668 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
