You Know the Science
You've spent a life time learning the science that comprises your field. You're not a programmer, but you've written programs to model, test, or simulate the science. You expect the results to be accurate.
How Well Do You Know the Code?
Exclusive of the algorithms involved, software can have many pitfalls including:
- Errors signaled but not caught properly (e.g., overflow / underflow, divide by zero),
- Units conversions and mismatches (e.g., meters <-> feet),
- Incorrect or improperly documented constants,
- Round off in data manipulation,
- Global variables that are shared in unexpected ways
We have a wide selection of analysis tools for most of the popular scientific programming languages including FORTRAN, C, and C++:
- Test Coverage - how much of your code is actually exercised by your test suite?
- Source Code Search Engine - quickly search your whole (multi-language) code base at a lexical, not a text string, level,
- Smart Differencer - compare two files at the lexical, not text string, level,
- Metrics - report popular complexity metrics for your code base,
- Clone Detection - finds copies and near copies in your code and help you refactor into common code.