[cfe-dev] Exclude protobuf generated files

Manuel Klimek klimek at google.com
Wed Nov 26 02:16:40 PST 2014

/subscribe - I'm very interested in false positives in protocol buffer code

On Wed Nov 26 2014 at 8:08:51 AM Ted Kremenek <kremenek at apple.com> wrote:

> Hi George,
> I think the mitigation strategy for reducing false positives right now
> from the analyzer are the two techniques you mention.
> You can use __clang_analyzer__ but it really is a heavy hammer, and
> essentially precludes the analyzer from doing anything useful with those
> files.  If you are using the analyzer from within Xcode, you also need to
> make sure that the source can still parse (during analysis) after you add
> the __clang_analyzer__ guards or Xcode will abort the analysis of your
> project because it encountered source files that can't parse.
> Specifically with respect to protocol buffer code, could you file one or
> more PRs about representative issues you are seeing?  There's a possibility
> they could be pruned using heuristics within the analyzer.  If protocol
> buffer code tends to be generated idiomatically with specific file
> extensions, e.g. .pb.cc, we could possibly use even that to tune some
> heuristics on the analyzer to avoid false positives.
> Cheers,
> Ted
> > On Nov 25, 2014, at 12:49 PM, George Kollias <georgioskollias at gmail.com>
> wrote:
> >
> > Hello all,
> > Recently, we started using the clang static analyzer in our project
> > and it's already very helpful. One thing we'd like to improve though
> > is to reduce the number of the warnings/bugs we get about protobuf
> > generated files -*.pb.cc- (we're not really interested in them). What
> > would be the best way to avoid analysing them without affecting
> > heavily the rest of the code? E.g. would it be better to avoid
> > generating these files completely or it'd be better to wrap each of
> > these files in #ifndef __clang_analyzer__/#endif blocks?
> >
> > Thank you,
> > George
> > _______________________________________________
> > cfe-dev mailing list
> > cfe-dev at cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20141126/9919940f/attachment.html>

More information about the cfe-dev mailing list