CSM - Courseware Student Module Touch Points

Part of PLAT-2436, this is an attempt to discover where the StudentModule model and table are read/written, for the purpose of crafting a solution to update its primary key from a signed int to an unsigned bigint.

Direct access to the StudentModule

Selects

• class_dashboard aggregates or selects across many StudentModule rows in single queries in these places:
  • https://github.com/edx/edx-platform/blob/master/lms/djangoapps/class_dashboard/dashboard_data.py#L37
  • https://github.com/edx/edx-platform/blob/master/lms/djangoapps/class_dashboard/dashboard_data.py#L80
  • https://github.com/edx/edx-platform/blob/master/lms/djangoapps/class_dashboard/dashboard_data.py#L109
  • https://github.com/edx/edx-platform/blob/master/lms/djangoapps/class_dashboard/dashboard_data.py#L430
  • https://github.com/edx/edx-platform/blob/master/lms/djangoapps/class_dashboard/dashboard_data.py#L494
• xblock_utils uses direct aggregate SELECT SQL executed here:
  • https://github.com/edx/edx-platform/blob/master/openedx/core/lib/xblock_utils/__init__.py#L271
  • Comment: "Print out a histogram of grades on a given problem in staff member debug info."
• course_blocks
  
  • Gets a single user/course/block row: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/course_blocks/utils.py#L25
• courseware
  
  • model_data
    
    • Gets multiple user/course/block rows: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/model_data.py#L952
    • Get on a single row: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/model_data.py#L1030

  • user_state_client

    • Get on multiple user/course/block rows using a chunked_filter: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/user_state_client.py#L81
    • Paginated iterator over all StudentModule objects for a given XBlock locator, pulling back rows in chunks: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/user_state_client.py#L414
    • Paginated iterator over all StudentModule objects for a given course, pulling back rows in chunks: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/user_state_client.py#L428
  • views
    • Gets all rows for a student/usage key/course (to then get their history): https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/views/views.py#L1200
• grades
  
  • Mgmt command recalculate_subsection_grades pulls all rows for a learner/course/subsection modified within a certain time: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/grades/management/commands/recalculate_subsection_grades.py#L64
• instructor
  
  • reset_student_attempts gets a single student/course/module state key: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/instructor/enrollment.py#L292
• instructor_analytics
  
  • list_problem_responses gets all answers to a single problem for a course, optionally in chunks: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/instructor_analytics/basic.py#L436
• instructor_task
  
  • _get_modules_to_update fetches all rows for a course / list of usage keys / optional user, also does a get_or_create:
• The admin site

Writes

• instructor
  • reset_student_attempts overwrites 'attempts' using save(): https://github.com/edx/edx-platform/blob/master/lms/djangoapps/instructor/enrollment.py#L337
• course_blocks
  
  • Modifies learner state: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/course_blocks/transformers/library_content.py#L101
• courseware
  • model_data
    
    • get_or_create for an individual row, plus a get, plus an update: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/model_data.py#L993
  • user_state_client
    
    • get_or_create in a loop, with potential updates using save() : https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/user_state_client.py#L246 
• instructor_task
  
  •  _get_modules_to_update does a get_or_create in a loop: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/instructor_task/tasks_helper/module_state.py#L427
• xmodule
  
  • lti_2_util's set_user_module_score publishes an event that claims to update StudentModule state, but I'm not sure what the code path for that is: https://github.com/edx/edx-platform/blob/master/common/lib/xmodule/xmodule/lti_2_util.py#L258
• The admin site

Deletes

• instructor
  
  • reset_student_attempts can optionally delete StudentModule rows recursively: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/instructor/enrollment.py#L299
• The admin site

Caches

• courseware
  • I'm unclear on how many of the caches and K/V stores in courseware/model_data.py access or store StudentModule data, ex: DjangoKeyValueStore & FieldDataCache: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/model_data.py#L680
  • courseware.masquerade MasqueradingKeyValueStore may also store state?: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/masquerade.py#L222

Indirect access to StudentModule

• LTI forces a publish of a user's score on a problem in these two places, which claims to eventually write to StudentModule:
  • https://github.com/edx/edx-platform/blob/master/common/lib/xmodule/xmodule/lti_2_util.py#L225
  • https://github.com/edx/edx-platform/blob/master/common/lib/xmodule/xmodule/lti_2_util.py#L237
• instructor_task _get_module_instance_for_task indirectly reads via DjangoKeyValueStore: https://github.com/edx/edx-platform/blob/master/lms/djangoapps/instructor_task/tasks_helper/module_state.py#L329
• I suspect there are several other indirect ways that StudentModule gets accessed that are hard to grep for! Expert opinions are needed here!

Special cases

• Tests that rely on StudentModuleFactory (many)
  
  • Obviously our fix must support or work around sqllite as well
• BokChoy tests use a cached schema that may need to be taken into consideration:
  • https://github.com/edx/edx-platform/blob/master/common/test/db_cache/bok_choy_schema_default.sql#L1681
• Tests that check query counts on these tables:
  • https://github.com/edx/edx-platform/blob/master/lms/djangoapps/courseware/tests/test_model_data.py

• StudentModuleHistoryExtended

  • Has event listeners an StudentModule for post_save and post_delete
  • Plus an "it's complicated" relationship with StudentModule.