courbet added a comment. I've ran this check on llvm. There are 0 instances of virtual function slicing (which is not surprising since these usually result in actual bugs) and 71 instances of member varaible slicing:
- 'FullSourceLoc' to 'SourceLocation': 57 - 'APSInt' to 'APInt': 5 - 'DeducedTemplateArgument' to 'TemplateArgument': 3 - 'SemaDiagnosticBuilder' to 'DiagnosticBuilder': 2 - 'RegHalf' to 'RegisterRef': 2 - 'PathDiagnosticRange' to 'SourceRange': 2 Most are harmless (but still true positives). The 'SemaDiagnosticBuilder' and 'APSInt' are actually interesting: llvm/llvm/tools/clang/lib/Sema/SemaExprObjC.cpp:3555 ``DiagnosticBuilder DiagB = …`` should be: ``SemaDiagnosticBuilder DiagB = …`` DiagB is then passed by reference to a function, so there really is no reason to slice it. AST/ExprConstant.cpp:3150 ``explicit APSInt(APInt I, bool isUnsigned = true)`` Here it’s easy to write: ``APSInt MyInt(MyOtherInt);`` and think it’s a copy, but it’s not (it just changed the signedness). ``APSInt MyInt(MyOtherInt.toAPInt());`` would make it clear what’s happening. http://reviews.llvm.org/D21992 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
