The new architecture memorizes settings (which we mostly did after reading the config so far) and also exposes the resulting file path as well as further settings.
This change is a prerequisite to define a dependency with Sprockets.
Previously, an exercise without visible files was used. While this worked for the test, it was semantically problematic, since the resulting submissions also lacked files.
We want to differentiate between those RfCs explicitly marked as solved, those potentially solved since the author reached the maximum score, and those being unsolved where the author has not yet reached the full score yet.
With this commit, we change the delivery method for the ACE editor from manually copied files to using yarn and webpack. As a side-change, we also modify how the mode is selected through JavaScript instead of Ruby.
Through webpack, the `modePath`, `themePath`, and `workerPath` are automatically determined and working as expected.
Closes#250
Previously, the filename was URL-encoded, thus each / was replaced with %2F. This caused issues with some Apache2 configuration, smartly mingling with the URL to either encode it a second time (resulting in %252F) or decoding it (generating a real /). However, for authenticated file downloads with the JWT, we hardly require a byte-by-byte matching. With these changes, the URL parameter is no longer URL-encoded, so that Apache2 won't break our implementation any longer.
Further, we use this opportunity to get rid of the unnecessary .json extension for those filename routes, simplifying the routes generated and doing some further cleanup.
With this commit, we refactor the overall score handling of CodeOcean. Previously, "Score" and "Submit" were two distinct actions, requiring users to confirm the LTI transmission of their score (after assessing their submission). This yielded many questions and was unnecessary, since LTI parameters are no longer expiring after each use. Therefore, we can now transmit the current grade on each score run with the very same LTI parameters. As a consequence, the LTI consumer gets a more detailed history of the scores, enabling further analytical insights.
For users, the previous "Submit" button got replaced with a notification that is shown as soon as the full score got reached. Then, learners can decide to "finalize" their work on the given exercise, which will initiate a redirect to a follow-up action (as defined in the RedirectBehavior). This RedirectBehavior has also been unified and simplified for better readability.
As part of this refactoring, we rephrased the notifications and UX workflow of a) the LTI transmission, b) the finalization of an exercise (measured by reaching the full score) and c) the deadline handling (on time, within grace period, too late). Those information are now separately shown, potentially resulting in multiple notifications. As a side effect, they are much better maintainable, and the LTI transmission is more decoupled from this notification handling.
Previously, it could happen under some (random) circumstances that a series of users got the same user name. This is caused by the usage of Forgery for generating a name. Forgery itself picks a random name from a list, and there is no check that two calls would indeed deliver two different names.
In order to solve the issue, we generate a truly unique email address for specs, so that the problem won't arise again.
Previously, the same runner could be used multiple times with different submissions simultaneously. This, however, yielded errors, for example when one submission time oud (causing the running to be deleted) while another submission was still executed.
Admin actions, such as the shell, can be still executed regardless of any other code execution.
Fixes CODEOCEAN-HG
FixesopenHPI/poseidon#423
* Remove deprecated options from environments
* Remove deprecation warnings for upcoming Rails 7.2
* Dump schema with new defaults
* Remove outdated (and erroneous) data attribute in view
* Resolve a `NoMethodError` for seeds_spec.rb
* Correct sorting in table
* Modify page when nested in exercises
* Fix links between pages
* Link from statistics page to programming_groups/index
* Link from submission page to programming_groups/<id>
* Allow filtering for exercise ID on ProgrammingGroup#index
* Add search fields for internal and external user id on pg/index
Since both projects are developed together and by the same team, we also want to have the same code structure and utility methods available in both projects. Therefore, this commit changes many files, but without a functional change.
* no lti session data anymore included in the function
* decided not to delete the pg_id here as it is handled in create_through_lti and implement
* the function is only called once without an exercise id so the values are now directly deleted there
* We want to identify a user that has triggered a testrun. Previously (in regular operation), only submission author who were regular users were able to start a testrun. Now, we want to prepare a future where submission authors are programming groups. Still, a testrun is triggered by an individual user and not a group.
* Further, this commit fixes some missing foreign key constrains.
