Lakeflow Declarative Pipelines release 2025.29
July 16 - 18, 2025
These features and improvements were released with the 2025.29 release of Declarative Pipelines.
Databricks Runtime versions used by this release
Channel:
- CURRENT (default): Databricks Runtime 16.1
- PREVIEW: Databricks Runtime 16.4
Because DLT channel releases follow a rolling upgrade process, channel upgrades are deployed to different regions at different times. Your release, including Databricks Runtime versions, might not be updated until a week or more after the initial release date. To find the Databricks Runtime version for a pipeline, see Runtime information.
New features and improvements
-
The Move Tables Between Lakeflow Declarative Pipelines feature is now generally available (GA). This enables users to move Materialized Views and Streaming Tables from one pipeline to another with a SQL command and a minor code adjustment. See Move tables between Lakeflow Declarative Pipelines
-
ALTER commands can now be run on MV/STs created by ETL pipelines for table comments, column comments, and RLS/CLM. As part of this feature, pipeline updates preserve (by column-name key) the values of fields that can be ALTERed by users rather than overwriting them. We’ll continue to respect explicit definitions in the source code, and anything not explicitly defined is preserved. The table below shows this behavior:
Column-level metadata:
Behavior | Existing state on Table/ST/MV | What’s defined on Pipeline source code or Create and Replace command | End state after Pipeline refresh or executing Create and Replace command | |
---|---|---|---|---|
Merge | We WILL NOT drop existing metadata on the Tables/STs/MVs before applying what’s defined in the Pipeline source code or Create and Replace command. | Column X with comment 1 | Column Y with Comment 2 | Column X with comment 1 Column Y with Comment 2 |
Column X with comment 1 | Column X with comment 3 Column Y with Comment 2 | Column X with comment 3 Column Y with Comment 2 |
Table-level metadata:
Behavior | Existing state on Table/ST/MV | What’s defined on Pipeline source code or Create and Replace command | End state after Pipeline refresh or executing Create and Replace command | |
---|---|---|---|---|
Merge | We WILL NOT drop existing metadata on the Tables/STs/MVs before applying what’s defined in the Pipeline source code or Create and Replace command. | Row filter 1 | Row filter 2 | Row filter 2 |
Row filter 1 | None | Row filter 1 |
You can continue to modify the source code as before. This merge approach preserves external ALTERs made to the table.
This new merge behavior introduces an important change. Before, pipeline updates deleted existing row filters or column masks if the pipeline definition did not include them. Now you must explicitly delete existing comments, row filters, or column masks, using the Catalog Explorer or the ALTER command.
For how to use the SQL commands, see ALTER MATERIALIZED VIEW and ALTER STREAMING TABLE.