Summary
This release adds list and list_effective methods to the workspace-level grants service alongside new fields for bundle deployments, disaster recovery, jobs, ML services, workspace repos, and model serving. Breaking changes include making the Postgres database spec role field required, removing the include_browse, browse_only, and external_secret_id fields from catalog secrets, and removing several ML and Postgres credential fields.
Summary generated by brickster.ai. For the full changelog and any code/binary attachments, follow the GitHub link above.
More from databricks/databricks-sdk-py
This release introduces a breaking change that removes the name field from the User service, replacing it with a new full_name field. Additionally, it adds visualization capabilities to the Genie service, SQL condition fields to job triggers, and various new configuration options across catalog, postgres, and serving services.
This release introduces cluster policy compliance updates, including a new method to cancel pending cluster enforcement and several new compliance-related fields, alongside support for AI runtime tasks in jobs and a new extra-large compute size for apps. Additionally, the workspace assets replication field for disaster recovery workspace sets is no longer required, and a new meta-harness user-agent dimension has been added.
This release adds new workspace-level service methods to manage Postgres data APIs and introduces the serverless compute ID field for Delta Live Tables pipeline creation, cloning, and editing. Additionally, it adds capacity reservation group attributes for Azure compute, endpoint ID fields for vector search indexes, and a Dynamics 365 connection type for Unity Catalog.
This release adds the type_overrides field to SyncedTableSpec and SyncedTableSyncedTableSpec, while making the resource_id field optional for bundle deployment operations. It also improves performance and stability on Spark Connect clusters by caching OIDC tokens, making WorkspaceClient.dbutils lazy, and falling back to the remote runtime implementation to prevent connection crashes.
This release introduces the new ai_search and bundle_deployments workspace-level services, alongside new configuration fields for catalog retention, connection settings, and pipeline clustering. It also contains a breaking change that removes the bundle package and its corresponding workspace-level service.
The Databricks Python SDK now supports the Vercel AI_AGENT environment variable as a secondary fallback for AI agent detection in the User-Agent header. Unrecognized fallback values will now pass through sanitized and length-capped at 64 characters instead of being coerced to an unknown agent status.