Creating java UDF for Spark SQL

[sm1274]

Hello, 

I have created a sample java UDF which masks few characters of a string. However I facing couple of issues when uploading and using it.

First I could only import it, which for now is OK. But when do the following,

create function udf_mask as 'baffleCrypto' using jar '/Workspace/Users/sumandra@baffle.io/baffle-databricks-udf-1.0-SNAPSHOT.jar' 
Throws an error 
[NOT_SUPPORTED_WITH_DB_SQL] CREATE FUNCTION is not supported on a SQL warehouse.

Is there an example on how to load a jar and then use it as external function. I followed the this guide 
https://docs.databricks.com/en/sql/language-manual/sql-ref-syntax-ddl-create-function.html

Even cut-n-paste errors out. 

Any ideas?

[Kaniz]

Hi @sm1274, The error message you received indicates that the CREATE FUNCTION statement is not supported on a Databricks SQL endpoint. This statement is the specific error message you're seeing indicating that you're trying to run the CREATE FUNCTION statement in the context of a SQL warehouse, which is not supported by Databricks.

It seems like you're trying to create a user-defined function (UDF) in Databricks using a JAR file. However, creating UDFs with JARs isn't supported in a Databricks SQL endpoint.

To create UDFs with JAR files, you need to use Databricks Runtime for ML (MLflow).