Lock the description store while a run writes to it [patch]

[matt-edmondson]

Fixes #164

What was wrong

Each process loads the whole store when it starts, and every save writes that whole in-memory copy back. When two runs overlapped, whichever saved last won:

• Two Scans over different folders lost one folder's descriptions.
• An Import or Configure made during a Scan was undone by the Scan's next per-image save.

The lock in Scan only serialized threads inside one process.

Change

This is option 1 from the issue, which the triage recommended.

• StoreLock. A new class that takes an exclusive lock file (FileShare.None) beside the store, at AppData.Path/ImageDescriber.lock. TryAcquire returns null while another run holds it.
• Which verbs lock. BaseVerb.WritesStore is overridden to true on Scan, Import and Configure. For those verbs, BaseVerb<T>.Run() does three things:
  • takes the lock for the whole run;
  • reloads the store from disk under the lock, so a copy that went stale after start-up, or during a Menu session, is never written back;
  • if the lock is held, prints "Another ImageDescriber run is updating the description store. Wait for it to finish and try again." and returns without running.
• Search, Stats and Export only read the store, so they don't take the lock.

Tests

StoreLockTests covers three things:

• A second acquire fails while the first is held, and succeeds after it is released.
• Exactly the three writing verbs report WritesStore.
• An Import started while another run holds the lock prints the busy message and never runs. The test can tell because its input file is missing, so a run would have printed "File not found".

With the lock wiring removed from BaseVerb, the two verb tests fail. The full suite passes: 90 passed, 1 skipped.

Note on #200

#200 ("Exit non-zero when … a verb fails") touches BaseVerb.Run too. The two branches merge cleanly; I checked with git merge-tree. Whichever lands second may want the busy refusal to set Failed = true, so a script sees the refused run as a failure.

🤖 Generated with Claude Code

https://claude.ai/code/session_01TTeaDNAaTEviJbNWuEi5hP

---

Generated by Claude Code

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
91.3% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud