Lecturer: Pietro Berkes, <pietro.berkes_AT_ googlemail _DOT_ com>

Download the cheat sheets with a summary of the tools and methods discussed in the class!

Slides: class slides

Exercises:

• carpentry_exercises.pdf (exercise sheet)
• carpentry_exercises.zip (source code)
• Exercise solutions (available on Tuesday afternoon)

The Live Demo TDD code, implementing the k-means algorithm: k_mean.zip

Outline of the lecture (2h):

1. The agile development cycle, scientific-style
2. Unit testing
   • unittest
   • testing with numpy arrays
   • what to test and how for scientific programming
3. Debugging
   • pdb
4. Profiling
   • timeit
   • cProfile
5. Coverage
6. Continuous integrations

Many live demos, music and dances!

Two comprehensive software carpentry courses:

sofware-carpentry.org

Tiziano's course at the Bernstein center in Berlin

The classic book about software carpentry: The pragmatic programmer

PEP8, the official Python style guide

pylint , tool to check coding standards

pyflakes, passive code checker (only errors, not style)

pep8, script to check PEP8 coding standard

unittest, the standard Python test framework

nose, an alternative framework that simplifies writing tests and allows for extension and customization of test experience. There is a useful option for running a test coverage analysis: nosetests –with-coverage

py.test, a less common alternative

doctest, write tests inside of docstrings

mock, the most popular tool for creating mock objects for testing.

pdb, the standard python debugger

winpdb, a graphical interface for pdb. It's platform-independent despite its name.

DDD (DataDisplayDebugger), graphical general-purpose debugger.

Python IDEs often have a built-in debugger.

cProfile, the batteries included Python profiler

timeit, built-in Python module to measure the execution time of small code parts

Tools to visualize profiling results:

RunSnakeRun, a wxPython-based visualization of profiling results – very intuitive!

PyCallGraph

GProf2Dot

KCacheGrind