sberg added a comment. A random data point from trying this patch on the LibreOffice code base:
- a little over 100 cases that are easily identified as false positives (many of the form "if (p) *p = ...") - two or three cases that looked suspicious on first glance but turned out to be false positives too (but where the code apparently benefits from cleaning up) - two cases of false positives that are awkward to silence (via "!= nullptr") in a way compatible with older C++ standards, as they are of the form "if (bool * p = ...)", and can be rewritten to "if (bool * p = ...; p != nullptr)" only for recent C++ - one false positive in system-provided /usr/include/qt5/QtCore/qstring.h - no true positives found https://reviews.llvm.org/D45601 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits