This commit introduces several improvements to database migration
scripts to enhance compatibility between MySQL and PostgreSQL, ensure
case-sensitive behavior where needed, and improve migration safety and
correctness. The changes include dynamic SQL generation based on the
database engine, improved transaction handling, and adjustments to
field types and adapters for better cross-database support.
Database compatibility and case sensitivity improvements:
- Migration scripts in split_modulestore_django and learning_sequences
now dynamically generate SQL statements for altering column case
sensitivity and uniqueness based on whether the database is MySQL or
PostgreSQL, ensuring correct behavior across both backends.
- common/djangoapps/split_modulestore_django/migrations/0001_initial.py
- openedx/core/djangoapps/content/learning_sequences/migrations/0001_initial.py
- The courseware.fields module now checks for "postgresql" in the
database engine string instead of a specific backend name, improving
compatibility with different PostgreSQL drivers.
- lms/djangoapps/courseware/fields.py
- The 0011_csm_id_bigint migration in courseware now supports both MySQL
and PostgreSQL for altering column types, with specific SQL for each
backend.
- lms/djangoapps/courseware/migrations/0011_csm_id_bigint.py
- The 0009_readd_facebook_url migration in course_overviews now
introspects the table structure using backend-specific SQL for MySQL
and PostgreSQL, ensuring correct detection of existing fields.
- openedx/core/djangoapps/content/course_overviews/migrations/0009_readd_facebook_url.py
Migration safety and correctness:
- Service user creation and deletion in the commerce app is now wrapped
in atomic transactions to ensure database consistency.
- lms/djangoapps/commerce/migrations/0001_data__add_ecommerce_service_user.py
- The move_overrides_to_edx_when migration in courseware now specifies
a no-op reverse migration, preventing accidental data loss on migration
rollback.
- lms/djangoapps/courseware/migrations/0008_move_idde_to_edx_when.py
Adapter registration and code cleanup:
- The common_initialization app now registers custom adapters for
CourseLocator and related classes in psycopg2 when using PostgreSQL,
ensuring proper serialization of these types.
- openedx/core/djangoapps/common_initialization/apps.py
- Minor code cleanup and formatting improvements in migration files,
including import order and field formatting for readability.
- lms/djangoapps/grades/migrations/0015_historicalpersistentsubsectiongradeoverride.py
- Returns top parent key instead of boolean in upstream info api
- Adds edited_on raw time in course outline api
- Adds has_changes to course details api
This PR ensures that the instructor "Problem Responses" report in Open edX includes all student responses to all problems under any selected block, including those that are nested or randomized (such as those from legacy library_content blocks). Previously, the report could miss responses to problems that were not directly visible to the instructor or admin user generating the report, especially in courses using randomized content blocks or deep nesting.
In courses that use randomized content (e.g., legacy library_content blocks) or have deeply nested structures, the instructor dashboard’s problem response report was incomplete. It only included responses to problems visible in the block tree for the user generating the report (typically the admin or instructor). As a result, responses to problems served randomly to students, or problems nested in containers, were omitted from the CSV export. This led to inaccurate reporting and made it difficult for instructors to audit all student answers.
The openedx-learning repo was recently refactored to consolidate its
authoring apps into a single openedx_content app:
https://github.com/openedx/openedx-platform/pull/37924
This commit makes the following changes to accommodate this:
- Bumps the openedx-learning version to 0.31.0 to get the changes.
- Creates new migrations in content_libraries, contentstore, and
modulestore_migrator to foreign key references to authoring apps
to point to the new openedx_content app. This is done without
actually making database changes, since the openedx_content app
models are taking over the existing tables that the authoring apps
once pointed to.
- Creates new squashed migrations in these apps that create these
foreign keys to reference openedx_content app models from the start.
The full rationale for how and why this was done is in the following
openedx-learning ADR:
https://github.com/openedx/openedx-learning/blob/main/docs/decisions/0020-merge-authoring-apps-into-openedx-content.rst
These migrations should run fine from either a from-scratch scenario
(i.e. a new install or CI), or when upgrading from an Ulmo-or-later
database state. If you have a database state that comes from the middle
of the Ulmo development cycle (e.g. October 2025), you may encounter
migration errors in content_libraries, contentstore, or
modulestore_migrator, with an error message complaining about missing
tables. If you receive this message, run the following command:
python manage.py lms migrate openedx_content 0001
Then try to run the migrations for the app that failed. Repeat if
necessary for multiple apps.
Required upgrade of Django.
Commit generated by workflow `openedx/openedx-platform/.github/workflows/upgrade-one-python-dependency.yml@refs/heads/master`
The caret version pinning change updated @edx/paragon from 2.6.4 to
2.7.0, which caused Icon component IDs to change from "Icon2" to
"Icon1" in snapshot tests.
This is a cosmetic change with no functional impact. The Icon
component generates unique IDs using a module-level counter in
src/utils/newId.js:
let lastId = 0;
const newId = (prefix = 'id') => {
lastId += 1;
return `${prefix}${lastId}`;
};
The ID values depend on module import order, which changed slightly
between versions. The visual rendering and accessibility features
(screen reader text) are unaffected.
References:
- Paragon Icon component: https://github.com/edx/paragon/blob/v2.7.0/src/Icon/index.jsx
- Paragon newId utility: https://github.com/edx/paragon/blob/v2.7.0/src/utils/newId.js
- Version comparison: https://github.com/edx/paragon/compare/v2.6.4...v2.7.0
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Change package.json to use caret (^) pinning for flexible version
resolution. Packages >=1.0.0 pin to major version, packages <1.0.0
pin to minor version.
The following packages remain exact-pinned due to compatibility issues:
- redux (3.7.2) and redux-thunk (2.2.0): Newer versions of redux-thunk
(2.3+) require redux@^4 as a peer dependency. Since the codebase uses
redux 3.x, allowing redux-thunk to upgrade would cause peer dependency
conflicts and potential runtime issues.
- @edx/frontend-component-cookie-policy-banner (2.2.0): Newer versions
(2.6.0+) depend on @openedx/paragon@21.x which requires PNG file loaders
and uses SCSS files incompatible with the current webpack/sass-loader
configuration.
- bootstrap (4.0.0): Newer versions (4.6.x) use a `deprecate` mixin in
their SCSS that the current sass compilation setup doesn't support.
- jasmine-core@2.6.4: Newer 2.x versions (2.99+) enforce stricter
afterEach placement rules that break edx-ui-toolkit's ajax-helpers.js
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
User retirement failed when CourseOverview records were missing, causing CourseOverview.DoesNotExist exceptions
in bulk email signal handler. These records were missing when the course was deleted.
Solution:
Add exception handling in force_optout_all signal handler:
- Wrapped Optout.objects.get_or_create() in try/except block
- Log warning and skip enrollment when CourseOverview is missing.
- Add GET /api/instructor/v2/courses/{course_id}/group_configurations
- Add GET /api/instructor/v2/courses/{course_id}/group_configurations/{id}
- Create shared constants module for course groups
Add a new endpoint to get the unit extensions for a course. Update edx-when from 3.0.0 to 3.1.0 in order to get data necessary for response.
---------
Co-authored-by: Daniel Wong <danieleduardo.wongfa@wgu.edu>
I want to start using coding assistants and play around with content in
the CLAUDE.md file but I don't want git to complain about it or check it
in for everyone just yet.
Sphinx tries to explore all the defined objects when trying to generate
docs from the code. When it tries to introspect the QuerySet object, it
results in that object trying to query the database which we don't
really want to setup for a docs build as it's not relevant and fairly
expensive to do. To circumvent this issue, we add a new filter to the
docs build that essentially skips further introspection of QuerySet
objects.
* refactor: moved remaining feature dicts settings into top-level settings.
* refactor: moved remaining feature dicts settings into top-level settings.
* fix: fixed the test files
* fix: fixed tehe pylint errors
* fix: fixation of the cms ci failure
* fix: fixed remaining feature settings for cms
* fix: added fix for requirements
* fix: added fix for lms tests
* fix: resolved the test views issue
* fix: configured views file and test_views
* fix: fixed lint errors and assertion issues
* fix: added fix for base url issue in test view
* fix: added fix for base_url and assertion issue
* fix: added configurations for base utl fix
* fix: handled none issue for mfe config
* fix: corrected override settings in test views
* fix: added getattr defensive technique for view settings
* fix: reverted views and test_views file
* fix: added settings in views file
* fix: added with patch within functions in test view
* fix: rearranged the features in default_legacy_config
* fix: fixing the tests with clearing cache
* fix: reverted test views to verify the CI check
* fix: added cache clear in mfe config test
* fix: fixed the patch toggles to override settings
* fix: fixed the lint errors
* fix: changed patch toggle to override settings
