Loop Analysis and Transformation towards STL Algorithms

Loops and iterations have always been a traditional error source in programming. “Off-by-one” errors, where an iteration is executed once too often or once too few, top the most common program errors. C++ provides STL algorithms to simplify the most common tasks accomplished using loops, helping to avoid these kinds of errors. Unfortunately, STL algorithms are not as commonly used as they could be.
This master thesis describes the development of an Eclipse C++ Development Tools (CDT) plug-in, which encourages and supports developers to use STL algorithms. The plug-in provides semi-automatic recognition and transformation of compatible loops to corresponding STL algorithms. To achieve this, tree pattern matching algorithms are applied to the processed abstract syntax trees (AST). The final plug-in features transformation of for each and find/find if -compatible for and while loops to equiv- alent STL algorithm function calls. The respective loop body is transformed into a corresponding functor. The user can select either a C++0x lambda expression, a TR1 bind expression or a C++98 bind1st/bind2nd expression.
Based on the foundation created with this project, there exist multiple extension possibilities. Additional algorithms, such as generate and transform, would greatly increase the number of use cases served. Furthermore, introducing trivial explicit function and explicit functor class functor transformations can increase the plug-in’s flexibility even further. Lastly, the tree pattern matching engine, with all its benefits, should definitely be provided to the Eclipse end users. Various applications, from tree pattern search masks to semi-automatic pattern definition based on existing codes, could support programmers in all of their daily tasks and challenges.