Contributing

Contributions are welcome, and are greatly appreciated! Every little bit helps, and credit will be given. While at its core Tabbycat is a software project, you do not need to know how to code or use Git in order to help. We welcome feedback and ideas based on your tabbing experience and appreciate suggestions or proposals for how to improve the wording, translation, and design of our interface and documentation.

Feel free to join our Facebook group if you have any questions about how to get started.

Feedback and ideas

These can be added as issues in the GitHub repository; posts in our Facebook group; or as an email to the developers.

Bug reports

Please report bugs by opening a new issue in our GitHub repository. It is most helpful if you can include:

  • How Tabbycat was installed (on Heroku, locally on macOS, etc.)

  • Any details about your tournament and setup that might be helpful in troubleshooting

  • Detailed steps for how to reproduce the bug

Getting started with development

  • To easily test your changes to Tabbycat you probably want a working local install (without using Docker)

  • Please submit pull requests for features and bug fixes against develop (but not master).

  • We broadly use the git-flow workflow.

  • We use Django’s testing tools — adding unit tests to new features is greatly appreciated

    • A number of our tests use Selenium and ChromeDriver to simulate in-browser functionality. They will fail if you do not have the Chrome browser and ChromeDriver installed.

  • A number of extra dependencies are required for running tests, linting, and serving the documentation. These can be installed with:

    $ pip install -r 'config/requirements_development.txt'
    
  • Our package.json provides a convenience command that runs a standard set of development tools simultaneously, such as the Django server and the automatic recompilation with live injecting of javascript and CSS. Once you have set USE_WEBPACK_SERVER=True in your settings_local.py you can then run this with:

    $ npm run serve
    

Generating test data

There are management commands to help developers quickly generate data for use in testing, including results and feedback. A list of all commands can be found from dj help, but the most useful in this context are:

  • dj importtournament ( minimal8team | australs24team | bp88team ), which imports participant data for the 8-team (minimal8team), 24-team Australs (australs24team) and 88-team BP (bp88team) demonstration tournaments respectively.

  • dj simulaterounds ROUND [ROUND ROUND ...], which simulates all of the rounds specified, generating a draw, an adjudicator allocation and a complete set of random results (but not feedback).

  • dj generatefeedback ROUND [ROUND ROUND ...], which randomly generates feedback for all existing debates in the specified rounds.

  • dj generateresults ROUND [ROUND ROUND ...], which randomly generates results for all existing debates in the specified rounds. (You don’t need to run this if you ran simulaterounds, because that already does it.)

Rounds can be specified by sequence number (seq) or abbreviation. You can find more information about each of them by adding --help after the command name.

Database schema changes

When adding new features, it may be necessary to modify the database schema to support these new additions. After the changes are made, the migration files made by python manage.py makemigrations must also be committed. The migration files should also contain methods fill the new fields based on existing data if possible.

Fixture files (found under data/fixtures/) may also need to be updated, which can be done by running the migrate_fixtures.py script under a unmigrated database, then committing the result.

$ python data/migrate_fixtures.py develop (your branch)

Style guide

For the front end interface design there is a style guide available at “/style/” once a tournament has been setup.

For python code, we use flake8 to check for a non-strict series of style rules. Warnings will trigger a Travis CI build to fail. The entire codebase can be checked by using:

$ flake8 .

For stylesheets, we use stylelint. The relevant code can be checked by using:

$ npm run lint-sass

For javascript, we use eslint to enforce the standardJS style and the standard recommendation of the vue plugin for eslint. The relevant code can be checked by using:

$ npm run lint-vue

Versioning convention

Our convention is to increment the minor version whenever we add new functionality, and to increment the major version whenever:

  • the database can’t be migrated forwards using python manage.py migrate --no-input, or

  • there is a major change to how the tournament workflow goes, or

  • we make some other change that is, in our opinion, significant enough to warrant a milestone.

We write data migrations to allow existing systems to be upgraded easily. However, we don’t always support backward database migrations. Our expectation is that long-lived installations keep up with our latest version.

One day, we hope to have a public API in place to facilitate the integration with other debating tournament software, like registration or adjudicator feedback systems. If and when that happens, we’ll probably revise this convention to be more in line with Semantic Versioning.

Starting from version 0.7.0, we use cat breeds as the code names for major versions.

Documentation

Documentation is created using Sphinx and hosted at Read The Docs. Pushes to develop will update the latest documentation set, while pushes to master will update the stable documentation set.

To preview the documentation locally, install the development dependencies and then start the server:

$ sphinx-autobuild docs docs/_build/html --port 7999

You should then be able to preview the docs at 127.0.0.1:7999.

Project structure

  • bin contains a number of convenience scripts for starting/stopping Docker, and the webserver/asset pipeline.

  • data contains the sample data sets and fixtures used to setup demo tournaments and in automated tests respectively

  • docs contains our document source files and images (although some are linked from the root directory)

  • tabbycat is the main directory containing the Django project
    • locale contains translation strings for shared templates (others are in respective app directories)

    • templates contains shared html templates, stylesheets, javascript source files, and Vue.js components/mixins.

    • utils contains shared utilities

    • All other folders are the Django apps that contain specific views, models, and templates for functions such as draw generation/display, or recording results. Each has sub-folders for tests and templates.

  • In the root directory there are a number of files defining our python and javascript dependencies, core configuration files, and key documents like the README

Internationalization/Localization

The gettext framework is used to enable the translation of strings in Python files and Django templates. Backend in this context signifies these types of files.

The backend’s translation files can be updated from the tabbycat directory using one or more of the supporting language codes (see settings.py):

$ dj makemessages -l es

To do more than one language, just specify -l multiple times, _e.g._ -les -lar.

These can then be compiled using:

$ dj compilemessages -l es

As it stands Heroku needs the .mo files pre-compiled (see issue in Heroku Python buildpack, so these are committed to Git. Note that the English (en) language files should not be compiled; their sole purpose is to provide a source language for Crowdin.

Strings defined in Vue files must similarily be marked with gettext but must be added manually to tabbycat/locale/LANGUAGE_CODE/djangojs.po, for each language supported. These can then compiled to javascript bundles using:

$ dj compilemessages -l es        # or whichever language(s) you want to update
$ dj compilejsi18n -l es

These are then also committed to git to save users needing to run compilejsi18n during setup. The resulting files are then bundled as part of the npm build task. Updating these translations in development (live-reload) requires the use of the cp-i18n npm task.

Release checklist

  1. Check that all migrations have been generated and committed into Git

  2. Merge translations from the Crowdin pull request and compile messages

  3. Bump version number in docs/conf.py

  4. Bump version number and (if applicable) codename in tabbycat/settings/core.py

  5. Update the main CHANGELOG.rst file (including release date)

  6. Check the major current deployment options, including:
    1. The deploy_heroku.py script

    2. The Tabbykitten version

    3. Docker (macOS, Windows 10*) and Docker Toolbox (Windows 10 Home) methods

    4. Using Bash and Powershell on Windows

    5. Using Terminal on macOS (at least test out a fresh install of the npm/pip dependencies)

  7. Check that the last Travis CI build passed and run the full local test suite (this will include the Selenium tests that are not on Travis)

  8. Shift remaining issues from the Github Milestone

  9. Create and finish the release branch as per git-flow

  10. Ensure the tag is correct (vX.Y.Z) and published to GitHub

  11. Back-merge master to the kitten branch

  12. Back-merge develop to the in-progress feature branches

  13. Issue a formal release with change notes on GitHub

  14. Post change notes on the Facebook page/group