Abstract
A method is disclosed for carrying detector-generated defect reports from a source case store into a target work-item store without duplicating a defect, recording a recurrence that arrives as a new report after closure as a regression rather than absorbing it. Each source record carries two keys: a report key, chosen by each writer for its own re-run idempotency in several incompatible shapes, and a content key, a content hash identifying the defect itself. Only the content key is matched across stores. The target store carries a unique index on tenant and content key that is partial on two predicates: rows without a content key are excluded, so the index needs no backfill and a keyless case can still be opened; and closed rows are excluded, so a recurrence's new record cannot collide with the closed one. A pure planner links a case to a record sharing its content key, treats a match on a record in its finished-status set as a regression and opens a new record citing it, and otherwise opens a record. Creating actions pass an admission ladder of windowed ceilings evaluated one decision at a time; a refused case is deferred by writing nothing, so it stays unpaired and is re-evaluated next poll. A one-to-one pairing relation, unique on both sides, doubles as a detector of duplicates in the source store, resolved by a merge that moves history onto the oldest record and archives the newer one without marking it resolved.
Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Assuncao, gustavo matthew, "Report-Identity vs Bug-Identity Split for Detector-Driven Intake: Closed-Row-Excluded Partial Unique Index, Row-Absent Deferral and Pairing-Detected Source Duplicates with Archive-Only Merge", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/13200