The motivation for this change is performance. The forums UI code gets
the list of users for each role and renders the staff label based on
those lists. The list for the staff role is expensive to compute because
there is no index on the is_staff attribute, and we cannot create one
because the User model is built into django.
Users with is_staff=True are still assigned the Moderator role upon
enrolling in a course, so this change will have no practical effect
except that a user who is granted staff privileges after enrolling in a
course will have to be made a Moderator in order for their posts to be
labeled.
Additionally, the UI did not use the list of users with the Student
role, so that list has been removed as well.
Avoid recomputing course module information for every thread, which
should dramatically improve the performance of high-percentile latency
queries.
JIRA: FOR-250
The _DISCUSSIONINFO global was originally used as a cache, but has since
lost that capability and is therefore just harmful. This is a precursor
to more refactoring that will improve the performance of the forums and
may itself provide some performance improvement because it separates the
computation done by two functions that each previously computed the
entirety of _DISCUSSIONINFO.
CommentClientError now has sane subclasses that are meaningfully
distinct, and each subclass is handled appropriately. Errors raised by
the requests library are no longer handled by turning them into
CommentClientErrors, since there is no meaningful handling we can do,
and this way we will get more visibility into why errors are occurring.
Also, HTTP status codes from the comments service indicating client
error are correctly passed through to the client.
Logging the duration of each request will allow us to determine
whether there is a significant difference in the latency reported by
the comments service and that observed by the LMS. Each request will
be assigned a unique identifier to allow correlation of the reported
latency on each end.
Previously, authentication was done using a URL parameter, which would
appear in various logs. Now, authentication is done more appropriately
with an HTTP header. Note that this requires cs_comments_service commit
cf39aabdd160176ebf206ca19d3ee030161a0b47 or later.
Previously, AJAX calls would return 400, and page views and attempts
to load inline discussions would return 404 if communication with the
comments service failed. Now such problems cause a 500 status code.
Features coming down the pipe will want to be able to:
* Refer to enrollments before they are actually activated (approval step).
* See what courses a user used to be enrolled in for when they re-enroll in
the same course, or a different run of that course.
* Have different "modes" of enrolling in a course, representing things like
honor certificate enrollment, auditing (no certs), etc.
This change adds an is_active flag and mode (with default being "honor").
The commit is only as large as it is because many parts of the codebase were
manipulating enrollments by adding and removing CourseEnrollment objects
directly. It was necessary to create classmethods on CourseEnrollment to
encapsulate this functionality and then port everything over to using them.
The migration to add columns has been tested on a prod replica, and seems to be
fine for running on a live system with single digit millions of rows of
enrollments.
