Debug UDFs using VSCode extension

SeyedA
New Contributor

I am trying to debug my python script using Databricks VSCode extension. I am using udf and pandas_udf in my script. Everything works fine except when the execution gets to the udf and pandas_udf usages. It then complains that "SparkContext or SparkSession should be created first.". I did some research and looks like SparkContext is not supported in Databricks Connect (read https://docs.databricks.com/en/dev-tools/databricks-connect/python/limitations.html). On the other hand I also found https://docs.databricks.com/en/dev-tools/databricks-connect/python/udf.html that says you can use UDFs with Databricks Connect. So I am confused. If you cannot debug UDFs with vscode extensions, how do you guys typically do this? Thanks for your help.

Retired_mod
Esteemed Contributor III

Hi @SeyedA, To resolve this, first, ensure your SparkSession is properly initialized in your script. Be aware of the limitations of Databricks Connect, which might affect UDFs, and consider running UDFs locally in a simple Spark environment for debugging. If VSCode becomes too challenging, Databricks Notebooks offer a more integrated environment. 

If you have any specific parts of your code that you’d like to share, I can help you troubleshoot further. How does that sound?