AaronBallman wrote: > > I'd like to understand the motivation for the tool a bit better. IMO, the > > goal should be to deprecate serialized diagnostics because we are now able > > to emit diagnostics to SARIF instead, and the whole point to SARIF is to be > > a machine-readable interchange format for diagnostic information to be > > shared between tools. So if we want to promote SARIF for sharing > > diagnostics between tools, it seems like we don't really need a tool to > > deserialized diagnostics we've serialized. > > CC @cjdb for awareness, as he's been working on SARIF support. > > The goal of these tools is to eventually be able to obtain clang diagnostics > as structural data. Say, I want the compiler to emit warnings and learn about > the number of warnings I will get. For us, this is something that's worth > putting into a database so that we can nudge relevant teams to make changes.
Okay, that makes sense to me. Yeah, I would recommend using SARIF for that. > I didn't know that we plan to deprecate serialized diagnostics. I would have > done differently if I had known about SARIF. I didn't mean to oversell it, I don't think we have firm plans to deprecate serialized diagnostics, just that it seems like something we're going to want to do at some point so we don't need to carry around multiple implementations of machine-readable diagnostic information. https://github.com/llvm/llvm-project/pull/118522 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
