Software

• UnitTest in the standard library (http://docs.python.org/lib/module-unittest.html)

• PyUnit at http://pyunit.sourceforge.net

• [http://www.garethrees.org/2001/12/04/python-coverage/ StatementCoverage] This module runs your code, then produces a report on how many statements were executed, and which ones were not. Use it to ensure your unit tests test everything.

• DataTest at http://formencode.org/docs/DataTest/README.html

• McCabe-like Python Cyclomatic Complexity analysis tools are available at http://journyx.com/curt/complexity.html. They're written in Perl, but read and analyze only Python code. Complexity is bad, this will help you simplify code - especially code you didn't write.

• [http://www.python.org/pypi/zope.testing zope.testing] provides a powerful test runner that supports test discovery and a wide range of options to control how tests are run and results reported.

Best Practices

Discussion

Sprouted out by http://formencode.org/docs/DataTest/TODO.html.

What I need is a layered test system like

• test suit
  • with fast/normal/detailed mode
  • with known failing tests excluded
• test package to group related tests with preset parameters
• individual test
  • with options like log details
  • with ability to flex parameters, extend options etc.
• test utilities
  • fuzzy difference
  • different logging/reporting/visualization helpers
  • with output capture capability

-- MikeRovner DateTime(2004-02-27T19:25:32Z)