Tutorial¶
The following sections outline some important topics in well managed, robust scientific software. These topics are not the only things that one could include in such a list, but they are common ones.
The implementations here are Python specific: the packaging, pytest for testing, sphinx for documentation, etc. But the notion of these topics are universal. You should use conventional packaging methods for the language you’re using. You should be automating your tests. You should have well managed documentation. Your command-line interfaces and logging behavior should follow established conventions. You should benchmark and profile the performance and memory footprint of your code.
These things will not make your scientific software correct. But doing these things are a big part of making your scientific software easy to use and easy to maintain.