Summary
Expose the internal scan cap consistently and prevent a non-authoritative count from looking like complete success.
Current v1.38.0 evidence / reproduction
find QueryCommandRunner --all --limit 3 --json emits only result rows and does not reveal the 250,000-line scan cap.
find QueryCommandRunner --all --count --json can return scan_truncated:true and authoritative_count:false with exit 0.
Impact
CI and audit tools can treat a lower bound as a complete repository count.
Suggested direction / acceptance criteria
- Emit terminal scan metadata for both row and count modes.
- Add
--require-authoritative or a documented partial-result exit status.
- Report the applied cap, scanned amount, and a recovery/continuation action.
Dogfooding candidate IDs: Q030, Q031. Verified against cdidx v1.38.0 at commit 2919bfc9b1b83cfa7ae1d964845d0f9ffca002d1.
Summary
Expose the internal scan cap consistently and prevent a non-authoritative count from looking like complete success.
Current v1.38.0 evidence / reproduction
find QueryCommandRunner --all --limit 3 --jsonemits only result rows and does not reveal the 250,000-line scan cap.find QueryCommandRunner --all --count --jsoncan returnscan_truncated:trueandauthoritative_count:falsewith exit 0.Impact
CI and audit tools can treat a lower bound as a complete repository count.
Suggested direction / acceptance criteria
--require-authoritativeor a documented partial-result exit status.Dogfooding candidate IDs: Q030, Q031. Verified against cdidx v1.38.0 at commit
2919bfc9b1b83cfa7ae1d964845d0f9ffca002d1.