Funcargs & other fun with pytest
13 likes14,047 views
The document discusses the features and advantages of using pytest for testing in Python, highlighting its support for various test styles, informative error reporting, and extensibility through plugins. It includes examples of using funcargs for dependency injection and shows its benefits in modular design and TDD practices. The author shares personal experiences with pytest in a forecasting application development context, noting its effectiveness and enjoyment over 18 months of use.
Funcargs & other fun with pytest
- 1. funcargs & other fun with pytest PyCon AU 2012 Brianna Laugher
- 2. pytest – all the good stuff ✔ all test styles: unittest, doctest, nose ✔ boilerplate-free style – natural asserts ✔ support for running tests distributed, boxed ✔ generate JUnit-style XML for Jenkins CI ✔ informative error reporting ✔ extensible plugins, useful builtins ✔ skip, xfail, custom markers ✔ parametrize, generate tests ✔ monkeypatch ✔ funcargs!
- 4. my background ● using pytest seriously for ~18 months at work ● really enjoying it! ● how it feels to use pytest: →
- 8. Graphical Forecast Editor ● client/server desktop application used by forecasters to generate the (text & graphical) forecasts ● pilot in VIC, funding to roll out nationally, ~5 year project, 20+ new devs/testers ● existing codebase of Python and C++ (used operationally in US) ● home-grown test suite and runner (via GUI)
- 12. extensible ● e.g. add in unittest assert methods
- 13. extensible ● e.g. add hook for winpdb
- 15. skip
- 16. xfail
- 17. custom markers
- 19. parametrize
- 21. generate tests ● defined once per module ● list/generate data to supply to 1+ tests ● way more flexible cf. py.test.mark.parametrize
- 24. monkeypatch
- 29. funcargs!
- 30. funcargs – idealised db example
- 31. funcargs – trivial django example
- 32. funcargs – trivial GAE examples
- 33. funcargs – trivial GAE examples
- 35. funcargs – why? ● dependency injection, aka inversion of control ● works a treat with TDD ● very natural with mocking ● encourages modular design
- 37. funcargs – a real DB example
- 38. funcargs – a real DB example
- 39. funcargs – to express prereq's ● for when a decorator is not powerful enough
- 41. funcargs – to hide monkeypatch :)
- 43. generate + funcarg - anypython
- 46. pytest – the future
- 50. test evolution - v1
- 51. test evolution - v2
- 52. test evolution – v3
- 53. test evolution - v4
- 55. pytest – all the good stuff ✔ all test styles: unittest, doctest, nose ✔ boilerplate-free style – natural asserts ✔ support for running tests distributed, boxed ✔ generate JUnit-style XML for Jenkins CI ✔ informative error reporting ✔ extensible plugins, useful builtins ✔ skip, xfail, custom markers ✔ parametrize, generate tests ✔ monkeypatch ✔ funcargs!
- 58. ● http://codespeak.net/mailman/listinfo/py-dev ● testing-in-python mailing list ● Stack Overflow ● lead developer Holger Krekel is very responsive
- 59. pytest.org
- 60. credits, thanks, contact Comic panels from comicallyvintage.tumblr.com Bureau of Meteorology ● In particular, Darius Powell introduced pytest to our project Code examples: gist.github.com/3386951 brianna@modernthings.org brianna.laugher.id.au / b.laugher@bom.gov.au @pfctdayelise