בתאריך 17 במאי 2014 22:39, "André Pönitz" <[email protected]> כתב: > > On Sat, May 17, 2014 at 08:54:21PM +0300, Jussi Pakkanen wrote: > > Some quick observations: > > > > - most false positives are cause by QTC_CHECK not marking its error > > function with QNORETURN > > But... QTC_CHECK always returns. Why should it have QNORETURN? > > > - there are several issues with Botan, which is a bit worrisome for a > > crypto library > > > > As an example of a found issue is a dereference of null pointer from > > qmljstypedescriptionreader.cpp, line 490: > > > > if (!ast || !ast->statement) { > > <- Assuming 'ast' is null > > addError(ast->colonToken, tr("Expected numeric literal after colon.")); > > <- Forming reference to null pointer > > > > That's about it. I hope you found this useful and/or informative. > > The first non-3rd-party issue (findtoolwindow.cpp) is genuine. Thanks. > A patch is on gerrit now. > > I already don't understand the second (branchdialog.cpp) m_model is > always non-null, it's created with new and assigned in the constructor > and never changed afterwards. > > Why does it say "Called C++ object pointer is null"?
This is already fixed. It warns since there was a function in which m_model was tested for null. > > Andre' > > _______________________________________________ > Qt-creator mailing list > [email protected] > http://lists.qt-project.org/mailman/listinfo/qt-creator
_______________________________________________ Qt-creator mailing list [email protected] http://lists.qt-project.org/mailman/listinfo/qt-creator
