aaron.ballman added a comment. In https://reviews.llvm.org/D30896#702913, @jbcoe wrote:
> I've played around with a few heuristics but it's still far too contentious > to have this check on by default and have it warn in places I want warnings. > Where should it go? Perhaps it should live as a private check rather than a public one? Alternatively, are you aware of any public coding standard rule that matches the behavior you want? We tend to be a little more relaxed about chattiness when it's part of a check for a public, relatively well-known coding standard because it guides the discussion about what constitutes a false positive. As a case in point, the CERT module has a check for MSC30-C, which prohibits using the `rand()` function. That would be way too contentious of a check normally except that it matches the behavior expected of a check against that CERT rule. I'm not certain we want a module named "chatty", or something like it, for checks that are over-eager to diagnose code, but that might be another option. It would require a community discussion outside of this code review. (Personally, I'd be opposed to such a module, but don't let that dissuade you from pitching the idea. I'm just one contributor and my mind can certainly be changed in the presence of good rationale.) Repository: rL LLVM https://reviews.llvm.org/D30896 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
