These removed tests can be restored once someone works out why they
were causing flakiness elsewhere and fix the problem.
Reverts tests added in:
dca50aacc3
The cache in the previous version of this code was unwittingly being
shared among all threads, and an occasional race condition would result
in the .children field of some XBlocks containing duplicate entries.
I tried to find other ways to keep the existing cache design and let it
be shared among all the threads (which would be more efficient), but I
couldn't find any clean way of doing it (and even then, this code was
not written with the intention of being used in a multi-threaded way).
So to keep the fix simple, I made the block data cache thread-local
instead of process-local. That eliminated the bug.
Technical details:
The big challenge with this code in the first place was due to the
parse_xml API of XBlocks, which instantiates the XBlock instance and
_then_ sets field data on it and returns it, as there is no mechanism
available to distinguish that from the case of instantiating an XBlock
(using previously loaded/cached field data) and then deliberately
changing its field values. In particular, parse_xml sets the 'children'
field just by calling self.children.append(...) but never explicitly
initializes self.children to [] first, so it's necessary for the field
data store to have a mechanism for setting self.children = [] when
parsing begins, but it's hard to know "when parsing begins" in general.
This is made more challenging since the BlockstoreFieldData design ties
field data changes to the XBlock instance itself (using a weakkey), but
it's impossible to get a reference to the XBlock instance before calling
parse_xml, and since the BlockstoreFieldData design uses a pass-through
approach where fields that aren't being actively changed are read from
the cache; since it doesn't know when children is being initialized vs.
being modified, it would sometimes pass-through and start one thread's
changes with the final result from another thread.
Anyhow, the bottom line is that avoiding unintentional multithreading
here solves the problem.
If we want the field data cache to be shared among threads, it might as
well be rewritten to use memcached and shared among all processes too.
That would be a very good performance boost but would take up a lot more
memory in memcached. Also the rewrite may be challenging due to the
aforementioned nuances of the XBlock parse_xml / construct_xblock /
add_node_as_child APIs. Perhaps modifying the runtime to use a
completely separate fielddata implementation for parsing vs. loading a
parsed+cached XBlock would do it.
Created sorting for email field
updated tests
Fixed sorting issue in registration form
Fixed sorting issue in registration form
Fixed sorting issue in registration form
Added missing items in env and updated order logic
Added missing items in env and updated order logic
ARCHBOM-1139: We were seeing about 150 queries to the django_site table
and over 200 to memcached just to load the login page. A lot of these
are via mako template rendering, but there are other sources, and this
will hopefully nip it in the bud.
Added more info to log in SSO request/response flow
Fixed django admin links on model's link fields which are broken due to django 2.2 upgrade.
ENT-2798
Fixed quality violations and unit test
Fix xsscommitlint violation
Fixed pylint violation
Without this PR, there is no [reasonable] way to get the following data
for any XBlocks in the new runtime; now there is :)
* index_dictionary: data about the block content, for search indexing
* student_view_data: data-only equivalent of student_view, for use in
custom UIs/mobile
* children: list of child IDs
* editable_children: list of child IDs in the same bundle (use case:
when showing an OLX editor you want to allow editing the OLX of
children in the same bundle but not linked children)
This adds some simple new python+REST APIs that can be used to create,
read, update, and delete "links" from a content library to other content
libraries.
One can use these links to import content (XBlocks) into a library
without copying the content.
Note that this feature was already fully supported by Blockstore and the
XBlock runtime; it's just that to use it prior to this required one to
use the (lower-level) Blockstore REST API directly to create the links.
Now there is a somewhat higher-level API built in to Studio, using
"content library" abstractions instead of Blockstore primitives.
* Remove email placeholder
The current placeholder is confusing to users and results in many trying
to login with their username. The change in this PR hopefully decreases that
confusion.
According to UX recommendations, we have decided to remove the email placeholder for the form. It seems placeholders tend to just confuse the users.
TNL-7053
The courseware_api view will use check_course_access - which now returns AccessResponse objects, and all other uses of check_course_access will now use check_course_access_with_redirect, which is a drop-in replacement for the original check_course_access implementation.
We also added a few new helpers to access_utils:
- check_public_access is a replacement for allow_public_access, which now returns AccessResponse objects
- check_enrollment checks if the learner is enrolled, and uses check_public_access to account for COURSE_ENABLE_UNENROLLED_ACCESS_FLAG
- check_survey checks whether there is a required survey that the learner must complete prior to accessing the course.
There are two new kinds of AccessError subclasses:
- SurveyRequiredAccessError
- EnrollmentRequiredAccessError
There was an issue where external LMS system (e.g. Canvas, Blackboard) that used Open edX LTI Provider calls had cookies blocked. This update fixes this issue by defining third-party cookies to have attributes of `Secure=True` and `SameSite=None`.
Details here: https://discuss.openedx.org/t/lti-xblock-and-samesite/759/5
(cherry picked from commit 28479a2966b87b16a25dbc96c19b6f5817d255de)
Prepare to upgrade to django-oauth-toolkit (module name oauth2_provider) 1.3.2 concurrently with the Django 2.2 upgrade (versions after 1.1.3 dropped support for Django 1.11). Key points:
* We can stop using a commit hash since versions 1.3.0 through 1.3.2 have been released since that commit.
* The validator in oauth_dispatch migration 0002 has been removed from DOT, but the whole table was removed in migration 0004 anyway. Newer DOT versions moved where redirect URI validation happens so it's no longer needed here.
* DOT 1.3.0 squashed its original 6 migrations and immediately deleted them. This required some changes in oauth_dispatch migrations 0006 and 0007 to deal with either numbering system.
When deploying to stage and production, we'll need to delete the history for oauth2_provider migrations 0002-0006 first (the new squashed 0001 migration has the same name as the original 0001). The deployment with DOT 1.3.2 will add a new 0002 migration which should then apply cleanly on the existing database state.
Should resolve https://openedx.atlassian.net/browse/BOM-1456 .
We now either pass in the relevant courseoverview or when creating the
enrollement we use the factory which automatically creates the relevant
CourseOverview object for testing purposes.
