- Revamped loading assemblies used inside UDFs (#180)
- Support for Vector UDFs (#127)
- Support for Grouped Map UDFs (#143)
- Resolve the signer mismatch issue (#186)
- Ability to launch a debugger from the worker (#150)
- Compatibility check for Microsoft.Spark.dll in the worker (#170)
- Update Apache.Arrow to v0.14.1 (#167)
- Support for RuntimeConfig in SparkSession (#184)
- Support for
Trigger
inDataStreamWriter
(#153) - The ability to use
--archives
option to deploy the worker binaries and assemblies that UDFs depend on (#187)
- Due to the changes in loading assemblies (#180), the Microsoft.Spark.Worker executable is no longer compatible with the previous versions.
- The
--class org.apache.spark.deploy.DotnetRunner
option inspark-submit
is changed to--class org.apache.spark.deploy.dotnet.DotnetRunner
to resolve the signer mismatch issue (#186).
The following table outlines the supported Spark versions along with the microsoft-spark JAR to use with:
Spark Version | microsoft-spark JAR |
---|---|
2.3.* | microsoft-spark-2.3.x-0.4.0.jar |
2.4.0 | microsoft-spark-2.4.x-0.4.0.jar |
2.4.1 | |
2.4.3 | |
2.4.2 | Not supported |