Automatic CDF: Row-level changes without table configuration

Automatic Change Data Feed (Auto CDF) computes row-level modifications during data reads by leveraging row tracking in Delta Lake and row lineage in Apache Iceberg v3 tables. Unlike the legacy version, it does not require manual table configuration and speeds up write operations by about 15%. ### Use cases Imagine you need to feed incremental pipelines or maintain audit trails without penalizing table write performance. - Process only changed rows in incremental ETL pipelines. - Sync modifications with external systems or caches. - Create audit trails for compliance and governance. ### How to use To read the change data feed in batch using SQL: sql SELECT * FROM table_changes('my_table', 0) What happens in this code: 1. The function reads the specified table. 2. The 0 parameter sets the starting trace version. 3. The engine computes changes at query time. To read in streaming using Python: ```python (spark.readStream .option("readChangeFeed", "true") .table("my_table") )