maxplanck-ie/parkour2 23.08.17 on GitHub

Full commit history: 0.3.9...23.08.17

Important News

We abandoned semantic versioning, from now on we will use dates. This is displayed on the login.html, righ above where users input their credentials.
parkour.env.sample has a new environmental variable (CSRF_TRUSTED_ORIGINS) that requires manual intervention/ copy to your deployments configuration. This way we are prepared for upcoming Django 4.2
Ever since our pre-relased 0.3.9 version, some seemingly redundant files corresponding to our frontend were removed. We did our best to fix any issues in this release. Yet, if you see a blank page on a fresh installation, please check the browser console (e.g. pressing F12) and file a new issue with the corresponding log message (e.g. pressing F12 and looking for the 'could not load **.js' or 'file not found' error messages.)

Updated all python requirements, but keeping on Django 3.2 to avoid an ongoing issue with sequencing depth miscalculation.
New CSRF_TRUSTED_ORIGINS option in misc/parkour.env, lists domains were application is deployed. You may use a wildcard to trust all subdomains. Be sure to add it, as in the file we provide: misc/parkour.env.sample. This will be mandatory once we update to Django 4.2
The docker image is now using Debian Bullseye as base, to match playwright system requirements.

Failed sequencing are now skipped from invoice XLS report. In the past staff users (sequencign team) needed to go request by request un-checking the 'sequenced' status on each.

new rules: coverage-html coverage-xml, run pytest with code coverage report(s).
new rule: playwright, to run end-to-end tests (frontend), for now 2 test were implemented.
The old test rule is now renamed as djtest, and it only runs the django unittests (functinal + integration).
added django-linear-migrations extension, to ease up fixing merge conflicts if we were to change models on different git branches.
added django-migration-linter with a new rule: lint-migras. This is now part of the test rule too, even though it's failing for 25 out of 38 migrations (see: incompatibilities for details, we have plenty of altering columns and a couple of missing default values on DB schema..)
Removed old debugging strategy of attaching term to docker container to work with PDB. It was too clunky, and we've been using werkzeug traceback interpreter anywayz.

New permission for users: 'Access as PI', grants them the ability to see all requests shared under their PrincipalInvestigator, whichever that is.
new rule: reload-code, sends the hung-up signal to green unicorn, gracefully reloading the wsgi config and the app code (details here) (note: development deployments run on werkzeug and auto-reload code via the docker bind mounts anyway!)
Fixtures (JSON) are in long format now, and a new django-admin custom command save_initial_data is available to share some of our data to other research facilities. The good old load_initial_data is there too.
Backup cronjobs are more separated in time now, to avoid a known issue with rsnapshot.
rule models now generates 3 PDF files, one is a simple A4 sheet for quick preview, and the other two are for printing posters in A1 size using multiple sheets in either A4 or A3 sizes.
Support for developers working on Windows is better now. Feel free to open an issue if you face further difficulties.

Submission of sample and libraries updated, we are hardcoding values for columns that are now hidden in frontend (e.g. concentration_method and RQN). Like always, custom parameters and its values can be passed in the free-text form inside Description or using file attachments to each request.