[SPARK-26244] Do not use case class as public API

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Done
• Affects Version/s: 3.0.0
• Component/s: Spark Core, SQL
• Labels: release-notes
• Assignee: Unassigned
• Reporter: Wenchen Fan
• Created: 02/Dec/18 02:48
• Updated: 06/Dec/18 14:40
• Resolved: 06/Dec/18 14:40

Description

It's a bad idea to use case class as public API, as it has a very wide surface. For example, the copy method, its fields, the companion object, etc.

I don't think it's expect to expose so many stuff to end users, and usually we only want to expose a few methods.

We should use a pure trait as public API, and use case class as an implementation, which should be private and hide from end users.

Changing class to interface is not binary compatible(but source compatible), so 3.0 is a good chance to do it.