Commit Graph
2 Commits
Author SHA1 Message Date
mika kunsandClaude Opus 4.8 f5d165baae fix(data): drop unique index on lists.name (allow duplicate list names)
The startup-race hardening added a global unique index on lists.name, but
duplicate list names are legitimate and the index broke 8 Worker tests that
seed same-named lists. The seeder race is already handled by the atomic
INSERT...WHERE NOT EXISTS, so the index is redundant. Keep the de-dup migration
step, remove the unique index from config, migration and model snapshot.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-09 10:15:42 +02:00
mika kunsandClaude Sonnet 4.6 7f1a14ab80 fix(data): harden FK pragma per-connection and seed concurrency
- Add SqliteForeignKeyInterceptor (DbConnectionInterceptor) registered via
  OnConfiguring so every IDbContextFactory-created context runs
  PRAGMA foreign_keys=ON, not only the MigrateAndConfigure context.
- DefaultListsSeeder: replace TOCTOU read-then-insert with atomic
  INSERT … SELECT … WHERE NOT EXISTS — one SQLite writer lock, no race.
- AppSettingsRepository.GetAsync: catch DbUpdateException on the
  get-or-create path and re-read so concurrent startup cannot throw.
- Migration 20260609000000_UniqueListName: de-duplicates empty list rows
  (startup-race leftovers) then adds a UNIQUE index on lists.name.
- ForeignKeyTests: verifies ON DELETE SET NULL (blocked_by_task_id) is
  enforced on a fresh DbContext with no manual PRAGMA call.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-09 10:05:41 +02:00