Home

Best Practices

Here are some dry lab best practices

Use version control (git)
Document code appropriately
- Auto-generate docs for complex software
- Per Google: "the best code is self-documenting"
  - Give sensible names to types and variables
- Use comments when appropriate
Use automated formatting
- Black, autopep8, rustfmt, clang-format, etc
- Automate the automated formatting: throw a pre-commit hook in .git/hooks
Use proper logging. Find an appropriate logging module for the environment. For bash script pipelines this may just mean echoing to a file. For Python use the logging package. We are doing science here, so having a hard record of everything that happens is very important. Logging can also be invaluable for troubleshooting. Some things to log:
- Write input and output filepaths to logs
- Write important parameters to logs
- Put the git branch and/or abbreviate commit hash in the logs. You can get this with sys calls:
  - $ git rev-parse --abbrev-ref HEAD
  - $ git log -1 --format=%h
- Put any important notes in logs
Write unit tests
- Unit tests are important. 100% code coverage may not be necessary, but use good judgment here
Use CI/CD
- Any standalone software should have CI/CD so that tests are run and the core functionality of the software is verified (does it work at all??) automatically. Automating deployment with CD means less manual work as well
Use relative filepaths