Automated testing

Automated tests are in tests which is divided in three sub-directories:

• unit: unit tests, test small amount of code in isolation. Good for algorithms, string generation, or generally everything that can be easily isolated. Shouldn’t do IO (database, files...) so they should be be fast.
• integration: integration tests, get a scratch-dir and an (in memory) database. Given the architecture of pyramid web apps, this is where most test will probably end-up. In particular, this is probably the place for view and model tests.
• end2end: End-to-end tests (or system tests). Those test a live server through a real browser using selenium. Good coverage but slow and hard to debug.

Running the test-suite

There are three ways to run the test suite:

• ./scripts/citest: this is the script that is used by the continuous integration server. It runs the whole test suite, generate a report in Xunit format and hide the browser window in a frame buffer (probably only works in linux).
• tox: will set-up a virtualenv for the test run and install test dependencies.
• py.test: you need to install the test dependencies yourself, most flexible.

For instance, to run only the end-to-end tests:

py.test tests/end2end/