Every branch shares the same dev-image and dev-image-base used for caching during the build image step.
On master, only tag commits are built and the image has the tag number

At the moment the staging server is kinda broken and the functional tests already ensure that the app is not completely broken

Dominik Seeger authored 6 years ago and robinwilliam.hundt committed 6 years ago

added instance export e2e test

Swapped Logout button and slot postion

Logout is now always a the far left

e2e tests are now using one browser instance per TestClass

Statistics endpoint now checks if there are submissiontypes available

Fixed race cond. in e2e test and bug in stats endpoint

Added function_tests to flake8 check

Added id "logout" to logout button

Using --cache-from in build step

Increased implicit wait for e2e test

Use postgres:9.6 is ci

* Also restructured the converter script so it handles
  mixed (gap and source code) export files as well
* Added a sad hack that enables using a base url

Fixed reverse query bug in Subscription model
Fixed bug in subscription view resulting in uncaught exception
Creating a subscription with a query/key/stage combination for which no assignments were available would result in an uncaught SubscriptionEnded exception and a 500 response to the client. Instead an error message with the status code 410_GONE is now sent.

Fixed reverse query bug in Subscription model

chnaged
type_query_mapper = {
	...
        SUBMISSION_TYPE_QUERY: 'type__title',
    }

to
type_query_mapper = {
        SUBMISSION_TYPE_QUERY: 'type__name',
    }

Refactored serializer id fields and camelCase names

To provide a uniform api and to save us from further work i've refactored the existing fields that used
camelCase names to use the names specified in the models (which are kebab-case).
Also everywhere where id's (whether normal or uuid ones) have been included in the serializers, the field names have been changed to 'pk' or '<model>_pk'. Pk will always link to the primary key of the model and will save us great pain should we decide to convert the pk's of more models to uuid's. Also we won't have to remebre a bunch of different ways of referring to the id for the frontend, it's always pk.
I also included the pk field in all modelserializers since this will be necessary for the frontend state management.

Frontend now expects pk fields and snake_case

Solution is highlighted / Desc. HTML is rendered

Frontend test is only manually run

Added vue-notification library

Inactivity detection preperly implementd

Client inactivity is now properly detected. A vuex plugin is used to store the time of the last commited mutation. This roughly equals the last user interaction.
If the users session is expired he will be redirected to the login page. Before that a dialog is displayed notifieng the user that they are about to be logged out.

Added created / of_tutor info to feedback comment

* Tests can now be executed from the importer
* All jobs run fine and produce correct results

* The mechanism proposed should work as follows:
  * Tutors can subscribe to certain submission categries (currently
    this includes exam, student or type specific submissions).
    If the set of submissions to corrent is small (student) all
    submissions of that category are reserved for that tutor.
  * A reviewer should also be able to subscribe other users (delegation)
  * A subscription contains assignments or creates them:
    * Only one assignment per user may be active.
    * No new assignments can be added to a subscription after it was
      created while another assignment is present for that subscription.
    * An assignment delegates a submission to a tutor.
    * An active assignment indicates that the tutor is working on that
      assignment
    * After an assignment was finished it is deleted (or archived).
* Upgraded to Django 2.0
* Closes #66, #53.

* The mechanism remains partially incomplete as the progress in
  !67-create-new-model-feedbackline-and-integrate-it is blocking
  progress. Several tests for the API endpoint need to be written
  including. More validation and constraints might have to be added.

This hopefully accelerates the pipelines as it uses
caching for pip modules and resticts building the
docker image only for the master branch (Which is
the only branch that is every deployed anyway)

* removing prospector as it is not compatible with latest version of flake8
* introducting a new CI job that runs flake8 and actually fails if there are any errors

* using gevnts for async requests handling of gunicorn
* serving the static frontend via gunicorn
* split requirements into dev and general
{

* Also added docstrings to classes and module in views.py
* Removed a logging statement from StudentApiView

- changed unittester from ./manage.py test to pytest in CI
- implemented tests for various views see previous commits
- coverage should now produce html output for test coverage

allowed to fail

- the forntend will now be part of the main image since its static
anyway
- The frontend has a testing stage
- reorganized .gitignore files
- docker-compose works as expected
 - serving api on 8000
 - serving frontend on 8080

Configured tes_coverage and test_pylint to change to backend dir

robinwilliam.hundt authored 7 years ago and Jan Maximilian Michal committed 7 years ago

Added Todo concerning the node_modules folder since the node modules are located at the Project root in development