Anish Mahto created SPARK-59786:
-----------------------------------
Summary: SCD1 Ignore-null support; Populate version map per event
row at ingest
Key: SPARK-59786
URL: https://issues.apache.org/jira/browse/SPARK-59786
Project: Spark
Issue Type: Sub-task
Components: Declarative Pipelines
Affects Versions: 4.4.0
Reporter: Anish Mahto
When processing an incoming microbatch, if ignore-null is enabled for the
AutoCDC flow, a non-null version map is populated for every upsert row. Delete
rows retain a null version map and are handled as row-wide authorship during
later reconciliation.
Per the version-map contract documented in `Scd1VersionMap`, the map contains
authorship information for every non-key user-data leaf in the target-aligned
row. This ticket stamps each upsert independently, using only that row’s values
and sequencing value.
This ticket does not reconcile microbatch version maps with existing auxiliary
or target-table state, nor does it reconcile multiple microbatch rows targeting
the same key.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
