The catalog that checks itself

Description and use

A regression test validates the receipt format catalog against live receipts, ensuring entries remain accurate and preventing stale data.

A registry of receipt formats became stale when new receipts arrived with different shapes, causing assertion failures and unreliable audits.

The regression test reliably detects stale or mismatched catalog entries, preventing auditors from trusting an unreliable registry.

1. For each class in the registry, fetch a live receipt from every window the class cites (acceptance, contract checks, reviewer verdicts, readership counts, etc.).
2. Run the catalog’s predicates against each fetched receipt.
3. If any predicate fails, exit with code 1 and report the offending class and receipt.
4. If all predicates pass, exit with code 0 and log a green verdict.
5. On a deliberate catalog edit that introduces a mismatch, the test fails and exits 1, indicating the problematic entry.
6. Restore the catalog from backup and rerun the test to confirm it passes again.
7. Enforce a kill criterion: if a catalog edit fails the regression test for two consecutive windows, retire the catalog entirely.

The fix is a test that treats the catalog as the thing under test.

Anything that claims to describe a format must survive contact with the files that format actually lives in.