Szelethus added inline comments.
================ Comment at: clang/lib/Frontend/CompilerInvocation.cpp:483-504 + if (!AnOpts.RawSilencedCheckersAndPackages.empty()) { + std::vector<StringRef> Checkers = + AnOpts.getRegisteredCheckers(/*IncludeExperimental=*/true); + std::vector<StringRef> Packages = + AnOpts.getRegisteredPackages(/*IncludeExperimental=*/true); + + SmallVector<StringRef, 16> CheckersAndPackages; ---------------- Szelethus wrote: > Szelethus wrote: > > The reason why I suggested validating this in CheckerRegistry is that > > CheckerRegistry is the only class knowing the actual list of checkers and > > packages, and is able to emit diagnostics before the analysis starts. This > > solution wouldn't work with plugin checkers/packages. > I don't see this being addressed actually? > > I think it would be totally fine to just omit the validation part as I said > earlier, the patch will be leaner, and cases in which we're using the > silencing of checkers are very exotic anyways. Also, we should probably compliment such validation by actually writing tests for plugins. I've been through that process once. It isn't fun. Really-really isn't :^) Let's just collectively agree to "forget" this :) CHANGES SINCE LAST ACTION https://reviews.llvm.org/D66042/new/ https://reviews.llvm.org/D66042 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits