- Production Website
- Changelog
- Contributing
- Getting Started
- Database & Assets
- Old Datatracker Branches
- Frontend Development
- Running Tests
This project is following the standard Git Feature Workflow development model. Learn about all the various steps of the development workflow, from creating a fork to submitting a pull request, in the Contributing guide.
Make sure to read the Styleguides section to ensure a cohesive code format across the project.
You can submit bug reports, enhancement and new feature requests in the discussions area. Accepted tickets will be converted to issues.
As outlined in the Contributing guide, you will first want to create a fork of the datatracker project in your personal GitHub account before cloning it.
Because of the extensive history of this project, cloning the datatracker project locally can take a long time / disk space. You can speed up the cloning process by limiting the history depth, for example (replace USERNAME with your GitHub username):
- To fetch only up to the 10 latest commits:
git clone --depth=10 https://github.com/USERNAME/datatracker.git
- To fetch only up to a specific date:
git clone --shallow-since=DATE https://github.com/USERNAME/datatracker.git
A beginning of a walkthrough of the datatracker models was prepared for the IAB AID workshop.
In order to simplify and reduce the time required for setup, a preconfigured docker environment is available.
Read the Docker Dev Environment guide to get started.
Nightly database dumps of the datatracker are available at
https://www.ietf.org/lib/dt/sprint/ietf_utf8.sql.gz
Note that this link is provided as reference only. To update the database in your dev environment to the latest version, you should instead run the
docker/cleandbscript!
We now use yarn to manage assets for the Datatracker, and vite/parcel to package them. yarn maintains its node packages under the .yarn directory.
The datatracker uses 2 different build systems, depending on the use case:
Pages will gradually be updated to Vue 3 components. These components are located under the /client directory.
Each Vue 3 app has its own sub-directory. For example, the agenda app is located under /client/agenda.
The datatracker makes use of the Django-Vite plugin to point to either the Vite.js server or the precompiled production files. The DJANGO_VITE_DEV_MODE flag, found in the ietf/settings_local.py file determines whether the Vite.js server is used or not.
In development mode, you must start the Vite.js development server, in addition to the usual Datatracker server:
yarn devAny changes made to the files under /client will automatically trigger a hot-reload of the modified components.
To generate production assets, run the build command:
yarn buildThis will create packages under ietf/static/dist-neue, which are then served by the Django development server, and which must be uploaded to the CDN.
The Datatracker includes these packages from the various Javascript and CSS files in ietf/static/js and ietf/static/css, respectively.
Static images are likewise in ietf/static/images.
Whenever changes are made to the files under ietf/static, you must re-run parcel to package them:
yarn legacy:buildThis will create packages under ietf/static/dist/ietf, which are then served by the Django development server, and which must be uploaded to the CDN.
The "new" datatracker uses Twitter Bootstrap for the UI.
Get familiar with https://getbootstrap.com/getting-started/ and use those UI elements, CSS classes, etc. instead of cooking up your own.
Some ground rules:
- Think hard before tweaking the bootstrap CSS, it will make it harder to upgrade to future releases.
- No
<style>tags in the HTML! Put CSS into the "morecss" block of a template instead. - CSS that is used by multiple templates goes into static/css/ietf.css or a new CSS file.
- Javascript that is only used on one template goes into the "js" block of that template.
- Javascript that is used by multiple templates goes into static/js/ietf.js or a new js file.
- Avoid CSS, HTML styling or Javascript in the python code!
If resources served over a CDN and/or with a high max-age don't have different URLs for different versions, then any component upgrade which is accompanied by a change in template functionality will have a long transition time during which the new pages are served with old components, with possible breakage. We want to avoid this.
The intention is that after a release has been checked out, but before it is deployed, the standard django collectstatic management command will be run, resulting in all static files being collected from their working directory location and placed in an appropriate location for serving via CDN. This location will have the datatracker release version as part of its URL, so that after the deployment of a new release, the CDN will be forced to fetch the appropriate static files for that release.
An important part of this is to set up the STATIC_ROOT and STATIC_URL settings appropriately. In 6.4.0, the setting is as follows in production mode:
STATIC_URL = "https://www.ietf.org/lib/dt/%s/"%__version__
STATIC_ROOT = CDN_ROOT + "/a/www/www6s/lib/dt/%s/"%__version__
The result is that all static files collected via the collectstatic command will be placed in a location served via CDN, with the release version being part of the URL.
In development mode, STATIC_URL is set to /static/, and Django's staticfiles infrastructure makes the static files available under that local URL root (unless you set settings.SERVE_CDN_FILES_LOCALLY_IN_DEV_MODE to False). It is not necessary to actually populate the static/ directory by running collectstatic in order for static files to be served when running ietf/manage.py runserver -- the runserver command has extra support for finding and serving static files without running collectstatic.
In order to work backwards from a file served in development mode to the location from which it is served, the mapping is as follows:
| Development URL | Working copy location |
|---|---|
| localhost:8000/static/ietf/* | ietf/static/ietf/* |
| localhost:8000/static/secr/* | ietf/secr/static/secr/* |
In order to make it easy to keep track of and upgrade external components, these are now handled by a tool called yarn via the configuration in package.json.
To add a new package, simply run (replace <package-name> with the NPM module name):
yarn add <package-name>Previous to this release, internal static files were located under static/, mixed together with the external components. They are now located under ietf/static/ietf/ and ietf/secr/static/secr, and will be collected for serving via CDN by the collectstatic command. Any static files associated with a particular app will be handled the same way (which means that all admin/ static files automatically will be handled correctly, too).
In order to make the template files refer to the correct versioned CDN URL (as given by the STATIC_URL root) all references to static files in the templates have been updated to use the static template tag when referring to static files. This will automatically result in both serving static files from the right place in development mode, and referring to the correct versioned URL in production mode and the simpler /static/ URLs in development mode.
During deployment, it is now necessary to run the management command:
ietf/manage.py collectstaticbefore activating a new release.
From a datatracker container, run the command:
./ietf/manage.py test --settings=settings_local_sqlitetestYou can limit the run to specific tests using the
--patternargument.
Frontend tests are done via Cypress. There're 2 different type of tests:
- Tests that test Vue pages / components and run natively without any external dependency.
- Tests that require a running datatracker instance to test against (usually legacy views).
Make sure you have Node.js 16.x or later installed on your machine.
To run the tests headlessly (command line mode):
yarn cypressTo run the tests visually (CANNOT run in docker):
yarn cypress:openIt can take a few seconds before the tests start or the GUI opens.
First, you need to start a datatracker instance (dev or prod), ideally from a docker container, exposing the 8000 port.
To run the tests headlessly (command line mode):
yarn cypress:legacyTo run the tests visually (CANNOT run in docker):
yarn cypress:legacy:openIt can take a few seconds before the tests start or the GUI opens.
To compare 2 different datatracker instances and look for diff, read the diff tool instructions.