Szelethus added a comment. In D58573#1422198 <https://reviews.llvm.org/D58573#1422198>, @NoQ wrote:
> Also, hmm, how about the following heuristics (looked at the positive in > `rtags` and thought about this): > > - if the field is public, don't warn (the structure seems to be used simply > as a convenient group of variables?) I disagree with this. If anything, public uninitialized fields should be reported the most. > - if all uninitialized fields are in all cases written into immediately after > the constructor ends, suppress the warning (i vaguely remember that something > similar was already proposed, but i might be wrong) I think we'd need dataflow analysis for that. We do suppress reports though where all fields of an object is uninitialized. Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D58573/new/ https://reviews.llvm.org/D58573 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
