The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
pycon	Ready	Preview	Feb 5, 2026 4:56pm

Summary

This PR adds a Django management command seed_review_data to generate test data for the review system. The command creates a conference, reviewers, submissions, grants, and associated review sessions with scores. This is useful for local development and testing of the review workflow.

Note: There's a typo in the PR title: "sedd" should be "seed".

Architecture & Design

✓ Good separation of concerns: The command is well-organized with private methods for each major entity type (_create_conference, _create_reviewers, _create_submissions, etc.), making the code readable and maintainable.

⚠️ Transaction handling: The command lacks transaction wrapping. If any part of the seeding process fails partway through, you could be left with partial/inconsistent data. Consider adding @transaction.atomic decorator to the handle method, similar to the fill_sponsor_data.py command at backend/sponsors/management/commands/fill_sponsor_data.py:15.

⚠️ Flush operation scope: The _flush() method at backend/reviews/management/commands/seed_review_data.py:115-118 deletes ALL review data globally, not just data created by this seed command. This could be dangerous if run on a database with real data. Consider:

• Adding a clear warning in the help text that --flush deletes ALL review data
• OR: Track seeded data (e.g., via a specific marker in conference code/name) and only delete that
• OR: Make it clear in documentation this is for local dev only

Testing & Coverage

❌ Missing tests: There are no tests for this management command. While seed/fixture commands are sometimes not tested, this command has complex logic that would benefit from at least basic smoke tests:

• Test that command runs without errors
• Verify the correct number of objects are created
• Test the --flush flag works correctly
• Test parameter variations (--submissions, --reviewers, --grants)

⚠️ Multi-tenant considerations: The command creates data for a specific conference but doesn't appear to have any tenant isolation issues since all created objects are properly scoped to the conference being created. However, the _flush() method breaks this isolation by deleting ALL reviews globally.

Error Handling

⚠️ No error handling: The command assumes all operations succeed. Consider handling potential failures:

• What if Conference.objects.create() fails due to unique constraint violations?
• What if get_or_create() calls fail for related objects (topics, languages, etc.)?
• The random.sample() call at backend/reviews/management/commands/seed_review_data.py:194 could fail if tag_names has fewer items than requested

⚠️ Silent failures: If conference creation fails or returns unexpected data, the subsequent operations may fail silently or create inconsistent state.

Performance

✓ Efficient bulk operations: The submission creation logic is reasonably efficient, creating users and submissions one at a time which is acceptable for a seeding command.

⚠️ N+1 queries potential: At backend/reviews/management/commands/seed_review_data.py:345-354, the review creation loops through submissions and reviewers, creating UserReview objects one at a time. For large datasets, consider using bulk_create():

reviews_to_create = []
for submission in submissions:
    for reviewer in selected_reviewers:
        reviews_to_create.append(UserReview(...))
UserReview.objects.bulk_create(reviews_to_create)

However, given this is a dev/test seeding script and the default is only 20 submissions, the current approach is acceptable.

✓ Reasonable defaults: The default values (20 submissions, 5 reviewers, 15 grants) are reasonable for local development.

Code Quality

⚠️ Hardcoded constants: The SCORE_OPTIONS, TALK_TITLES, and REVIEWER_NAMES at the top are good, but the speaker creation logic at backend/reviews/management/commands/seed_review_data.py:244 uses hardcoded UUID generation which might create conflicts if run multiple times without --flush. The uid prefix helps but isn't foolproof.

⚠️ LazyI18nString usage: The command correctly uses LazyI18nString for i18n fields, but only provides English translations. This is fine for seeding, but worth noting.

✓ Good output: The command provides helpful stdout messages showing what's being created, which is good UX for a management command.

Recommendations

1. High priority: Add @transaction.atomic to prevent partial data creation on errors
2. High priority: Fix the _flush() method to be safer (add warnings or scope to seeded data only)
3. Medium priority: Add basic tests to ensure the command works as expected
4. Medium priority: Add error handling for potential failures (unique constraints, etc.)
5. Low priority: Consider using bulk_create() for reviews if you expect to increase defaults
6. Low priority: Fix the typo in the PR title

Positive Notes

• Clear code organization with well-named helper methods
• Realistic test data (varied speaker levels, tags, multiple submissions per speaker)
• Flexible command-line arguments for customization
• Good feature: Creates speakers with multiple submissions to test that scenario

Codecov Report

❌ Patch coverage is 0% with 163 lines in your changes missing coverage. Please review.
✅ Project coverage is 92.70%. Comparing base (41c9e3e) to head (c117348).

@@            Coverage Diff             @@
##             main    #4559      +/-   ##
==========================================
- Coverage   94.16%   92.70%   -1.47%     
==========================================
  Files         353      354       +1     
  Lines       10347    10510     +163     
  Branches      761      780      +19     
==========================================
  Hits         9743     9743              
- Misses        502      665     +163     
  Partials      102      102