Development¶
Setting up a local copy of the Wagtail git repository is slightly more involved than running a release package of Wagtail, as it requires Node.js and npm for building JavaScript and CSS assets. (This is not required when running a release version, as the compiled assets are included in the release package.)
If you’re happy to develop on a local virtual machine, the docker-wagtail-develop and vagrant-wagtail-develop setup scripts are the fastest way to get up and running. They will provide you with a running instance of the Wagtail Bakery demo site, with the Wagtail and bakerydemo codebases available as shared folders for editing on your host machine.
You can also set up a cloud development environment that you can work with in a browser-based IDE using the gitpod-wagtail-develop project.
(Build scripts for other platforms would be very much welcomed - if you create one, please let us know via the Slack workspace!)
If you’d prefer to set up all the components manually, read on. These instructions assume that you’re familiar with using pip and virtual environments to manage Python packages.
Setting up the Wagtail codebase¶
The preferred way to install the correct version of Node is to use Fast Node Manager (fnm), which will always align the version with the supplied .nvmrc
file in the root of the project. To ensure you are running the correct version of Node, run fnm install
from the project root.
Alternatively, you can install Node.js directly, ensure you install the version as declared in the project’s root .nvmrc
file.
You will also need to install the libjpeg and zlib libraries, if you haven’t done so already - see Pillow’s platform-specific installation instructions.
Fork the the Wagtail codebase and clone the forked copy:
git clone https://github.com/username/wagtail.git
cd wagtail
With your preferred virtualenv activated, install the Wagtail package in development mode with the included testing and documentation dependencies:
pip install -e ."[testing,docs]" -U
Install the tool chain for building static assets:
npm ci
Compile the assets:
npm run build
Any Wagtail sites you start up in this virtualenv will now run against this development instance of Wagtail. We recommend using the Wagtail Bakery demo site as a basis for developing Wagtail. Keep in mind that the setup steps for a Wagtail site may include installing a release version of Wagtail, which will override the development version you’ve just set up. In this case, to install the local Wagtail development instance in your virtualenv for your Wagtail site:
pip install -e path/to/wagtail"[testing, docs]" -U
Here, path/to/wagtail
is the path to your local Wagtail copy.
Development on Windows¶
Documentation for development on Windows has some gaps and should be considered a work in progress. We recommend setting up on a local virtual machine using our already available scripts, docker-wagtail-develop or vagrant-wagtail-develop
If you are confident with Python and Node development on Windows and wish to proceed here are some helpful tips.
We recommend Chocolatey for managing packages in Windows. Once Chocolatey is installed you can then install the make
utility in order to run common build and development commands.
We use LF for our line endings. To effectively collaborate with other developers on different operating systems, use Git’s automatic CRLF handling by setting the core.autocrlf
config to true
:
git config --global core.autocrlf true
Testing¶
From the root of the Wagtail codebase, run the following command to run all the Python tests:
python runtests.py
Running only some of the tests¶
At the time of writing, Wagtail has well over 5000 tests, which takes a while to
run. You can run tests for only one part of Wagtail by passing in the path as
an argument to runtests.py
or tox
:
# Running in the current environment
python runtests.py wagtail
# Running in a specified Tox environment
tox -e py39-dj32-sqlite-noelasticsearch -- wagtail
# See a list of available Tox environments
tox -l
You can also run tests for individual TestCases by passing in the path as
an argument to runtests.py
# Running in the current environment
python runtests.py wagtail.tests.test_blocks.TestIntegerBlock
# Running in a specified Tox environment
tox -e py39-dj32-sqlite-noelasticsearch -- wagtail.tests.test_blocks.TestIntegerBlock
Running migrations for the test app models¶
You can create migrations for the test app by running the following from the Wagtail root.
django-admin makemigrations --settings=wagtail.test.settings
Testing against PostgreSQL¶
Note
In order to run these tests, you must install the required modules for PostgreSQL as described in Django’s Databases documentation.
By default, Wagtail tests against SQLite. You can switch to using PostgreSQL by
using the --postgres
argument:
python runtests.py --postgres
If you need to use a different user, password, host, or port, use the PGUSER
, PGPASSWORD
, PGHOST
, and PGPORT
environment variables respectively.
Testing against a different database¶
Note
In order to run these tests, you must install the required client libraries and modules for the given database as described in Django’s Databases documentation or the 3rd-party database backend’s documentation.
If you need to test against a different database, set the DATABASE_ENGINE
environment variable to the name of the Django database backend to test against:
DATABASE_ENGINE=django.db.backends.mysql python runtests.py
This will create a new database called test_wagtail
in MySQL and run
the tests against it.
If you need to use different connection settings, use the following environment variables which correspond to the respective keys within Django’s DATABASES settings dictionary:
DATABASE_ENGINE
DATABASE_NAME
DATABASE_PASSWORD
DATABASE_HOST
Note that for MySQL, this must be
127.0.0.1
rather thanlocalhost
if you need to connect using a TCP socket
DATABASE_PORT
It is also possible to set DATABASE_DRIVER
, which corresponds to the driver
value within OPTIONS
if an SQL Server engine is used.
Testing Elasticsearch¶
You can test Wagtail against Elasticsearch by passing the argument --elasticsearch7
or --elasticsearch8
(corresponding to the version of Elasticsearch you want to test against):
python runtests.py --elasticsearch8
Wagtail will attempt to connect to a local instance of Elasticsearch
(http://localhost:9200
) and use the index test_wagtail
.
If your Elasticsearch instance is located somewhere else, you can set the
ELASTICSEARCH_URL
environment variable to point to its location:
ELASTICSEARCH_URL=https://my-elasticsearch-instance:9200 python runtests.py --elasticsearch8
Unit tests for JavaScript¶
We use Jest for unit tests of client-side business logic or UI components. From the root of the Wagtail codebase, run the following command to run all the front-end unit tests:
npm run test:unit
Integration tests¶
Our end-to-end browser testing suite also uses Jest, combined with Puppeteer. We set this up to be installed separately so as not to increase the installation size of the existing Node tooling. To run the tests, you will need to install the dependencies and, in a separate terminal, run the test suite’s Django development server:
export DJANGO_SETTINGS_MODULE=wagtail.test.settings_ui
# Assumes the current environment contains a valid installation of Wagtail for local development.
./wagtail/test/manage.py migrate
./wagtail/test/manage.py createcachetable
DJANGO_SUPERUSER_EMAIL=admin@example.com DJANGO_SUPERUSER_USERNAME=admin DJANGO_SUPERUSER_PASSWORD=changeme ./wagtail/test/manage.py createsuperuser --noinput
./wagtail/test/manage.py runserver 0:8000
# In a separate terminal:
npm --prefix client/tests/integration install
npm run test:integration
Integration tests target http://127.0.0.1:8000
by default. Use the TEST_ORIGIN
environment variable to use a different port, or test a remote Wagtail instance: TEST_ORIGIN=http://127.0.0.1:9000 npm run test:integration
.
Browser and device support¶
Wagtail is meant to be used on a wide variety of devices and browsers. Supported browser / device versions include:
Browser |
Device/OS |
Version(s) |
---|---|---|
Mobile Safari |
iOS Phone |
Last 2: 16, 17 |
Mobile Safari |
iOS Tablet |
Last 2: 16, 17 |
Chrome |
Android |
Last 2 |
Chrome |
Desktop |
Last 2 |
MS Edge |
Windows |
Last 2 |
Firefox |
Desktop |
Latest |
Firefox ESR |
Desktop |
Latest |
Safari |
macOS |
Last 3: 15, 16, 17 |
We aim for Wagtail to work in those environments. Our development standards ensure that the site is usable on other browsers and will work on future browsers.
Unsupported browsers / devices include:
Browser |
Device/OS |
Version(s) |
---|---|---|
Stock browser |
Android |
All |
IE |
Desktop |
All |
Safari |
Windows |
All |
Accessibility targets¶
We want to make Wagtail accessible for users of a wide variety of assistive technologies. The specific standard we aim for is WCAG2.1, AA level. Here are specific assistive technologies we aim to test for, and ultimately support:
NVDA on Windows with Firefox ESR
VoiceOver on macOS with Safari
Windows Magnifier and macOS Zoom
Windows Speech Recognition and macOS Dictation
Windows High-contrast mode
We aim for Wagtail to work in those environments. Our development standards ensure that the site is usable with other assistive technologies. In practice, testing with assistive technology can be a daunting task that requires specialized training – here are tools we rely on to help identify accessibility issues, to use during development and code reviews:
@wordpress/jest-puppeteer-axe running Axe checks as part of integration tests.
Axe Chrome extension for more comprehensive automated tests of a given page.
Accessibility Insights for Web Chrome extension for semi-automated tests, and manual audits.
Known accessibility issues¶
Wagtail’s administration interface isn’t fully accessible at the moment. We actively work on fixing issues both as part of ongoing maintenance and bigger overhauls. To learn about known issues, check out:
The WCAG2.1 AA for CMS admin issues backlog.
The audit also states which parts of Wagtail have and haven’t been tested, how issues affect WCAG 2.1 compliance, and the likely impact on users.
Compiling static assets¶
All static assets such as JavaScript, CSS, images, and fonts for the Wagtail admin are compiled from their respective sources by Webpack. The compiled assets are not committed to the repository, and are compiled before packaging each new release. Compiled assets should not be submitted as part of a pull request.
To compile the assets, run:
npm run build
This must be done after every change to the source files. To watch the source files for changes and then automatically recompile the assets, run:
npm start
Using the pattern library¶
Wagtail’s UI component library is built with Storybook and django-pattern-library. To run it locally,
export DJANGO_SETTINGS_MODULE=wagtail.test.settings_ui
# Assumes the current environment contains a valid installation of Wagtail for local development.
./wagtail/test/manage.py migrate
./wagtail/test/manage.py createcachetable
./wagtail/test/manage.py runserver 0:8000
# In a separate terminal:
npm run storybook
The last command will start Storybook at http://localhost:6006/
. It will proxy specific requests to Django at http://localhost:8000
by default. Use the TEST_ORIGIN
environment variable to use a different port for Django: TEST_ORIGIN=http://localhost:9000 npm run storybook
.
Compiling the documentation¶
The Wagtail documentation is built by Sphinx. To install Sphinx and compile the documentation, run:
# Starting from the wagtail root directory:
# Install the documentation dependencies
pip install -e .[docs]
# or if using zsh as your shell:
# pip install -e '.[docs]' -U
# Compile the docs
cd docs/
make html
The compiled documentation will now be in docs/_build/html
.
Open this directory in a web browser to see it.
Python comes with a module that makes it very easy to preview static files in a web browser.
To start this simple server, run the following commands:
# Starting from the wagtail root directory:
cd docs/_build/html/
python -m http.server 8080
Now you can open http://localhost:8080/ in your web browser to see the compiled documentation.
Sphinx caches the built documentation to speed up subsequent compilations. Unfortunately, this cache also hides any warnings thrown by unmodified documentation source files. To clear the built HTML and start fresh, so you can see all warnings thrown when building the documentation, run:
# Starting from the wagtail root directory:
cd docs/
make clean
make html
Wagtail also provides a way for documentation to be compiled automatically on each change.
To do this, you can run the following command to see the changes automatically at localhost:4000
:
# Starting from the wagtail root directory:
cd docs/
make livehtml
Automatically lint and code format on commits¶
pre-commit is configured to automatically run code linting and formatting checks with every commit. To install pre-commit into your git hooks run:
pre-commit install
pre-commit should now run on every commit you make.
Using forks for installation¶
Sometimes it may be necessary to install Wagtail from a fork. For example your site depends on a bug fix that is currently waiting for review, and you cannot afford waiting for a new release.
The Wagtail release process includes steps for static asset building and translations updated which means you cannot update your requirements file to point a particular git commit in the main repository.
To install from your fork, from the root of your Wagtail git checkout (and assuming the tooling for building the static assets has previously been installed using npm install
), run:
python ./setup.py sdist
This will create a .tar.gz
package within dist/,
which can be installed with pip
.
For remote deployments, it’s usually most convenient to upload this to a public URL somewhere and place that URL in your project’s requirements in place of the standard wagtail
line.