Mining Discussions on Software Migration: A study of the Boost mailing list regarding C++ code evolution

• Pedro V. R. de Carvalho UnB
• Rodrigo Bonifácio UnB
• Walter Lucas UnB
• Alana Paula Barbosa Mota UnB

Resumo

Programming languages are evolving faster than ever before. New versions of mainstream programming languages (e.g., C++, Java, and JavaScript) are being released with increasing frequency, posing an elevated challenge for software developers as their systems are more easily affected by obsolescence. Software migration is far from trivial. Although there is literature on software migration methods and how developers deal with the software aging and obsolescence, little research exists on how developers perceive and are affected by rapid programming language evolution. To understand how C++ developers discuss these issues and the nature of their discussions, we mined the mailing lists of the Boost organization—one of the most important C++ open-source communities. We found that software migration is a significant concern for this community, with a lasting presence in their message boards. Furthermore, most discussions related to the challenges of the migration process, with many conflicting opinions on related matters, suggesting these issues are not easily solvable.

Palavras-chave: Data Mining, Machine Learning, Natural Language Processing, Software Migration, Modernization, Programming Language Evolution, Boost, C