Last year I described C++ as bearing the cautionary label ‘Here Be Dragons.’ And yet we’re all still writing C++ because it is the best programming language for the problems we face. In turn, we need a strategy to deal with the “dragons” that reside in large C++ code bases. The Dragon Book (my old compiler textbook) taught about a collection of tools to address the complexity of compiler design, and while our challenge is somewhat different, the approach remains the same. When a mere mortal programmer ventures forth to battle the complexity of large software systems in C++, they’re going to need some really good tools to help them. At Google, we’ve been building up a platform of such tools. I will introduce the platform and toolset, and show how to use them to write fast and correct C++ code, quickly and correctly.I will also give a peek into the future of the next generation of tools we’re working on and some of the really interesting changes to C++ that are coming in the next few years to help both programmers and these tools be ever more effective.
Andrey Karpov on 29 Mar 13OpenCV is a library of computer vision algorithms, picture processing algorithms, and general-purpose numerical algorithms. The library is written in C/C++ and is free both for academic and commercial use, as it is distributed under the BSD license. The time has come to check this library with the PVS-Studio code analyzer. OpenCV is a large library. It contains more than 2500 optimized algorithms and consists of more than 1 million code lines. Cyclomatic complexity of the most complex function cv::cvtColor() is 415. It is no wonder that we have found quite many errors and questionable fragments in its code. However, the size of the source code taken into account, we may call this library a high quality one.
