This article lists maintenance fixes and updates for supported DataForge versions.
DataForge Version 10.1
- April 21, 2026
- Fixed Unity Catalog source hub tables from being created with missing tbl properties
- Updated Rule Template expressions to reflect upstream attribute renames
- Fixed "missing watermark column" on custom refresh for Unity Catalog sources
- Fixed Unity Catalog source from "Delta Failed to Merge Fields" error during CDC
- Fixed Convert to Unity failing due to data type issues during CDC table creation
- Fixed Input Delete from failing on Unity Catalog sources when Data Profile is enabled
- Fixed issue with recalculate leaving uv_flag attributes as null values
- April 28, 2026
- Removed inactive channels from virtual output query generation
- [BEHAVIOR CHANGE] Hard Dependencies now use ingestion datetime rather than file last updated datetime to release workflow queue records
- Enabled file creation and download in Talos
- Fixed issue where new rule template leaves Cast Datatype as null rather than empty
- Fixed custom process race condition failures
- Fixed Source size listed on main sources page for Unity Catalog sources which now shows Ingot table size
- Added support for long-running custom loopbacks by attaching same job run id to both the Post Output process and Custom Ingestion process. Canceling the job run now cancels both processes together rather than leaving custom ingestion process in progress.
- [OPTIONAL CHANGE] Added support for two and three part naming in hub view names which allows you to define the catalog.schema.view_name in the source settings hub view name. This feature requires that you enable permissions on the DataForge-linked user in Databricks to manage views in the listed catalog and schema. This allows the views to be created and updated.
- Added system config parameter “fail-process-on-view-creation” with default = false. Changing this parameter determines whether the process is marked failed if the view is not able to be created or updated.
- Fixed Cleanup system-driven process to also delete inputs from source tables in addition to input metadata
Updated