Developer’s guide¶

This section contains information on how to perform various task and an overview of how our development infrastructure is set up.

Local development¶

These tasks are usually performed on an individual developer’s machine.

Run the test suite¶

The tox automation tool is used to run tests inside their own virtualenv. This way we can be sure that we know which packages are required to run the tests. By default tests are run in a Postgres database created by docker-compose. Other databases can be used by setting the DJANGO_DB_... environment variables. See Specifying the database.

$ ./tox.sh

By default, tox will run the test suite using the version of Python used when we deploy and will compile a local version of the documentation. The -e flag may be used to explicitly specify an environment to run. For example, to build only the documentation:

$ ./tox.sh -e doc

tox environments¶

The following tox environments are available.

py36: Run by default. Launch the test suite under Python 3.6. Generate a code-coverage report and display a summary coverage report.
doc: Run by default. Build documentation and write it to the build/doc/ directory.
flake8: Run by default. Check for code-style violations using the flake8 linter.
manage: Run manage.py management commands.

Run management commands¶

There are wrappers which allow running management commands either in a development environment (which mounts the work directory as a volume) or in the production image build (which must be built first). For example:

$ ./manage_development.sh gatherstats   # respects local changes

$ ./compose.sh production build         # build a production image
$ ./manage_production.sh                # uses contents of production image

Building the documentation¶

This documentation may be built using the “doc” tox environment.

Docker images¶

The application is deployed using Docker containers on the Google Container Engine.

Cloud infrastructure¶

This section provides a brief outline of cloud infrastructure for development.

Source control¶

The source code is hosted on GitHub at https://github.com/uisautomation/django-boilerplate. The repository has master set up to be writeable only via pull request. It is intended that local development happens in personal forks and is merged via pull request. The main rationale for this is a) it guards against accidentally git push-ing the wrong branch and b) it reduces the number of “dangling” branches in the main repository.

Unit tests¶

The project is set up on Travis CI to automatically run unit tests and build documentation on each commit to a branch and on each pull request.

Note

By logging into Travis CI via GitHub, you can enable Travis CI for your personal fork. This is highly recommended as you’ll get rapid feedback via email if you push a commit to a branch which does not pass the test suite.

In order to better match production, Travis CI is set up to run unit tests using the PostgreSQL database and not sqlite. If you only run unit tests locally with sqlite then it is possible that some tests may fail.

Code-coverage¶

Going to CodeCov, logging in with GitHub and adding the django-boilerplate repository will start code coverage reporting on pull-requests.

Documentation¶

Travis CI has been set up so that when the master branch is built, the documentation is deployed to https://uisautomation.github.io/django-boilerplate via GitHub pages. The UIS robot machine account’s personal token is set up in Travis via the GITHUB_TOKEN environment variable.

Code-style¶

The tox test runner will automatically check the code with flake8 to ensure PEP8 compliance. Sometimes, however, rules are made to be broken and so you may find yourself needing to use the noqa in-line comment mechanism to silence individual errors.

To run the flake8 tests manually, specify the tox environment:

$ ./tox.sh -e flake8

Documentation¶

This documentation is re-built on each commit to master by Travis and posted to GitHub pages at https://uisautomation.github.io/django-boilerplate/.