Implement an update pipeline that applies Rubin APDB replica update records to PPDB catalogs stored in HATS, so the HATS representation stays aligned with Observatory PPDB semantics without requiring full reimport for every change.
User Story
As a scientist or platform user querying PPDB through LSDB/HATS, I want that PPDB results in HATS remain scientifically correct, current, and comparable to Observatory-backed sources.
Scope
The initial implementation should support the update record types currently represented in APDB that dont touch solar system objects including:
- reassignment of DiaSource to DiaObject
- reassignment of DiaSource to SSObject - I suggest deleting these records for the time being, as we are not dealing with SS yet
- withdrawal of DiaSource
- withdrawal of DiaForcedSource
- closing DiaObject validity
- updating DiaObject nDiaSources
Acceptance Criteria
- A HATS update pipeline exists that consumes APDB update records and applies them in deterministic order.
- Applying the same ordered update record batch twice is safe or explicitly prevented with documented behavior.
- Updated HATS output passes validation against a reference BigQuery subset for representative cases.
- Publish procedure exists so that consumers can identify the latest promoted PPDB HATS version.
- Rollback procedure is documented for a bad update batch.
Implement an update pipeline that applies Rubin APDB replica update records to PPDB catalogs stored in HATS, so the HATS representation stays aligned with Observatory PPDB semantics without requiring full reimport for every change.
User Story
As a scientist or platform user querying PPDB through LSDB/HATS, I want that PPDB results in HATS remain scientifically correct, current, and comparable to Observatory-backed sources.
Scope
The initial implementation should support the update record types currently represented in APDB that dont touch solar system objects including:
Acceptance Criteria