Requested by Arvind:

regarding the fat jar, it would be VERY useful if your team builds it and publishes to Maven. The reason for this is because in Azure the most popular Spark deployment is Azure Databricks. Now, Databricks (in Azure and in AWS, but I am only looking at Azure) users are predominantly coding via. notebooks as opposed to submitting a traditional JAR to the cluster. Given this notebook interface and the user profiles for our customers, we have seen that if there are too many dependency JARs have to be added as libraries, it causes adoption friction and the customer would most likely not use the library. Hence, if you can release an uber JAR officially on Maven that would be really helpful to drive adoption of Spline with Azure Databricks.