Release History

Unreleased

Fix: Enum columns no longer raise TypeError: String.__init__() got an unexpected keyword argument '_enums'. Enum now resolves to a dedicated DatabricksEnumType instead of being adapted through the String colspec, keeping Enum validation and length inference while escaping literals the Databricks way (fixes #61)

Fix: Quote bind parameter names containing non-identifier characters (e.g. hyphens, backticks) so columns and parameters with special characters bind correctly (#60 by @msrathore-db)
Fix: Bind UUID values in canonical hyphenated form (#63 by @sreekanth-db, fixes #50)
Fix: Support pandas multi-row inserts with mixed-type columns via adaptive CAST, including bind names that contain escape-map characters, with an opt-out gate (#68, #69 by @msrathore-db)
Fix: Use public user_agent_entry connect parameter instead of deprecated _user_agent_entry to silence the deprecation warning from databricks-sql-connector >= 4.0.1 (#64 by @jprakash-db)
Fix: Promote Float(precision > 24) to DOUBLE in CREATE TABLE to preserve precision (#65 by @jayantsing-db)

Feature: Added pool_pre_ping support via do_ping() override to detect and recycle dead connections (#54 by @msrathore-db)
Fix: Pinned poetry version in CI workflows to fix build failures (#54 by @msrathore-db)

Feature: Added support for complex data types such as DatabricksArray and DatabricksMap [Private Preview] (#30 by @jprakash-db)

All the SQLAlchemy features from databricks-sql-connector>=4.0.0 have been moved to this databricks-sqlalchemy library
Support for SQLAlchemy v2 dialect is provided