>> I'm also still not convinced that you really need a //new// PathDiagnosticConsumer, rather than, say, improving the text diagnostics. > > I've posted a patch for clang-tidy, which adds code using this functionality, as an example: http://llvm-reviews.chandlerc.com/D2620 > > The problem is, that if we don't add a custom PathDiagnosticConsumer, we'd need to pass the checker name through clang's diagnostic system, which has no dedicated facility for this. We could add it, but it doesn't seem generally useful for other parts of clang.
Sorry, I meant improving the ClangDiagPathDiagConsumer to print the checker names (probably controlled by an -analyzer-config option, but possibly even on by default). The checker hierarchy is not a wonderful long-term solution, but it's what we've got, even for people using scan-build. http://llvm-reviews.chandlerc.com/D2556 _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
