[mnapoli]

Summary

Follow-up to #1156, stacked on it. Profiles hold the GitHub identity of users, so the old copy on the users table goes.

Changes

• Copies the identity of users who have no profile yet. Those are the users the old code signed up during the previous deploy.
• Adds a foreign key from users.github_id to profiles.github_id, so every user has a profile.
• Drops username, name, avatar_url, bio, location, website and github_synced_at from users.
• Removes the tests that only wrote to those columns: the copy migration's test (a user without a profile can no longer exist) and the old mention-backfill migration's test (it read users.username).

Validation

Local, on a throwaway Postgres:

• pest --parallel --ci --coverage --exactly=100.0: 4292 passed, 100% coverage.
• StressSeeder (STRESS_SCALE=0.001) and OperationsSeeder run with the foreign key: every user has a profile.
• Pint and Rector: no changes. PHPStan: the same 9 errors as on production.
• Deploy simulation: migrated to the previous PR, inserted a user the way the old code would (no profile), then ran this migration. The user got a profile with their identity, and the columns dropped.

Independent review

Reviewed by a separate read-only agent. Verdict on 3273dab0: PASS. Its one low finding is applied in 3750058: the Tom "wrong identity" test now only gets the GitHub ID wrong, and it fails if TomQuery stops checking it.

Limitations

• Deploy only once refactor: move the github identity of users to profiles #1156 is fully live: no web instances or queue workers left on older code. Older code still writes username and has no profiles, so sign-ups would fail on the foreign key.
• Pre-deploy check: this read-only query on production should return 0. Otherwise the copy hits the unique handle and the migration rolls back.

  select count(*) from users u
  where not exists (select 1 from profiles p where p.github_id = u.github_id)
    and exists (select 1 from profiles p where p.handle = u.username);

• No way back: the dropped columns can't come back without a new migration.

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}