Add a group-by-section

[AlexRedBlue]

For assignments (quizzes), added an option for the graph to group by section (violin style).

[IonMich]

ah, that's nice! I didn't know that this method can force refresh the cache.

[AlexRedBlue]

Useful if you have run an older version before. I think it was causing an issue with not adding the new button correctly. There's probably a way to do it dynamically too if there are future updates to detail.js for example.

[IonMich]

Thanks for the PR Alex! This is a useful UI element to add in the assignment detail page of the Bootstrap UI. Other than the minor get_section_in_course comment, everything else looks good!

[IonMich]

Here, can we use get_section_in_course from students/models.py ?

Students who belong in multiple sections of different courses can appear in the database. There are safeguards so that a student always belongs to only one section of a course during initial import, but the role of get_section_in_course is to raise an error if a student accidentally belongs in multiple sections in the same course.

[AlexRedBlue]

Good point. Could do something like this and catch the error too. Maybe even replace the print with an displayed error.

try:
    section = submission.student.get_section_in_course(assignment.course)
except ValueError as e:
    print(f"Error: {e}")
    section = None

[IonMich]

Yes that code should work. The print statement is also fine. I believe in the UI we are not properly displaying info anywhere for "unassigned section" / "multiple sections" students, so let's not display an error here. The chart UI in this PR already creates a violin for "Unassigned" students and that will be good enough indication to the TA that something needs fixing.

(I think the course detail page and/or the student detail page could show warnings for such situations, but that's probably separate from this PR.)

[AlexRedBlue]

I think everything seems okay now. I was seeing if I could dynamically update scripts tags with something like v={hash}, but I would need to do a bit more research into doing that properly.

[AlexRedBlue]

I think everything seems okay now. I was seeing if I could dynamically update scripts tags with something like v={hash}, but I would need to do a bit more research into doing that properly.

Perhaps all the script tags could be updated like so,

?v={{ ASSET_VERSION }}

Have GitHub Actions set an env var like ASSET_VERSION=${GITHUB_SHA} and use in templates via a context processor.

[IonMich]

I think everything seems okay now. I was seeing if I could dynamically update scripts tags with something like v={hash}, but I would need to do a bit more research into doing that properly.

Perhaps all the script tags could be updated like so,

?v={{ ASSET_VERSION }}

Have GitHub Actions set an env var like ASSET_VERSION=${GITHUB_SHA} and use in templates via a context processor.

Thank you Alex! I hadn't thought that this could be reliably automated easily - what you are suggesting would resolve the Ctrl+shift+R that I have to do after each git pull. I believe this asset generation though would create an extra commit for each push/merge to origin/main, right?

In order to avoid that extra commit we could have ASSET_VERSION be calculated dynamically at runtime from the git SHA (something like git rev-parse --short HEAD). But that would probably raise an error in machines without git (e.g. some of the department's Windows machines, where instructor_pilot can still be installed through Github Desktop). Anyway, feel free to create an issue and/or a new PR if you come up with something!

Merging this one!