erichkeane wrote: > This direction looks good to me, mostly I just have nits. > > The one major change I'd consider before landing is to find a way to avoid > the verbosity regression cor3ntin mentions for `Diag()` and friends. > Notational regressions within Sema itself may be as important as those of its > clients, as it's so much code.
I'd be in favor of just using a 'base' class for these and adding a few of the 'common' ones (`Diag`, `ASTContext`, `LangOpts`), and adding others if we found the need. `SourceManager` isn't really used enough in these sorts of things to be valuable, nor is PP or the FP Features. The OpenCL features might be worth adding, but only to the OpenCL breakout. https://github.com/llvm/llvm-project/pull/84184 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
