v0.17.0

• Disable downstream ucx check (#495). The downstream check for UCX has been disabled due to its current version being pinned to less than 0.17, rendering the checks unnecessary, and because the implementation of these checks exposes an issue with numerous UCX unit tests failing when used in conjunction with the latest Databricks SDK. As a result, the UCX check has been temporarily commented out to prevent failures, while other downstream checks, such as the one for remorph, continue to remain active, allowing for a more stable testing environment until the necessary updates can be made to UCX to address these issues and enable the reactivation of its downstream check.
• Explicitly test/support python 3.13 and 3.14 (#490). The acceptance workflow has been enhanced to support Python versions 3.13 and 3.14, in addition to existing versions, by updating environment variables and dependencies. The workflow now utilizes Hatch version 1.16.5, installed using the HATCH_VERSION environment variable, and the latest versions of actions/checkout and actions/setup-python, specifically version 6. The Python version used in the Install Python step has been updated to 3.14, and the types of pull requests that trigger this workflow have been modified to exclude ready_for_review events. Additionally, the condition to run the integration job has been simplified, and environment variables HATCH_VERBOSE and HATCH_VERSION have been added to the workflow. These changes aim to extend the CI/CD pipeline and project metadata to cover Python 3.13 and 3.14, with incidental updates to dependencies, including upgrading Hatch and GitHub actions, to improve the overall testing and installation process.
• Fixed dashboard serialisation: query is now queryLines on upstream REST API (#491). The Dataset class has been enhanced with improved serialization and deserialization functionality, allowing it to handle compatibility issues with dashboard APIs. The from_dict method now checks for both query and queryLines attributes in the input dictionary, using the query attribute directly and joining queryLines into a single string, while setting the query to None if neither attribute is found. This change enables the method to load datasets from various sources, including previously saved files, while maintaining compatibility. Additionally, the class's serialization and deserialization functionality has been thoroughly tested with the introduction of several new test methods, which verify its ability to correctly handle different query formats, including cases where the query is represented as queryLines or "query", and when the query is absent, ensuring accurate conversion between the object and dictionary representations.
• Limit sqlglot to releases earlier than 26.30.0 (#410). The library has undergone several updates to improve its functionality and robustness. The dependency on a key library has been updated to prevent the use of a version that introduced a breaking change, while also incorporating type-hinting fixes to accommodate newer versions of a static type checker. The Tile class has been enhanced with various changes, including input validation, improved position calculation, and better handling of potential issues, such as missing or invalid values. Additionally, numerous data classes have been modified to allow for optional values in various fields, enabling greater flexibility in the data models and facilitating the representation of incomplete or missing data. These changes aim to improve the overall reliability and adaptability of the library, allowing it to handle a wider range of scenarios and input data.
• Updated the minimum version of databricks-labs-pytester to 0.7.2 (#408). The minimum required version of a testing library has been updated to 0.7.2, which resolves an issue caused by a breaking change in the Databricks SDK version 0.51. This update ensures compatibility and prevents import failures when using older versions of the testing library with the new Databricks SDK. As a result, test dependencies must be at version 0.7.2 or later, although this change does not impact normal installs or new development environments, which will automatically use the latest version.

Contributors: @asnare, @JCZuurmond