https://bugs.kde.org/show_bug.cgi?id=507547
--- Comment #11 from Matthias Grimrath <[email protected]> --- You dare to go, were I wouldn't, by using allocators :D This, however, looks suspicious to me inline bool inContainer() { const auto &a = container()->get_allocator(); return this >= a.lastData && this < a.lastData + a.lastSize; so the pointer is compared against the allocated memory. However, a pointer somewhere into that allocated memory may not point to an actually used element. We saw QVector<> reserving some space at the beginning to be able to quickly insert elements at the top. And a STL implementation of std::vector<> is probably allowed to do the same. I don't know weather that leads to real-world problems. But I feel uneasy about it. I think a better approach would be to control insertion of new elements into 'm_lines' and update indices there, instead of hooking into QVector<> or std::vector<> by template magic. Should be the same performance wise. I'll work on a patch, having real code to show is probably better than sophisticated arguments :D -- You are receiving this mail because: You are watching all bug changes.
