[Implementation Question]: Should we override Equals and ToString?

[GoEddie]

Does it make sense to override these in spark objects and pass them back to the JVM (or provide our own implementation)?

Is there a preference for overriding or not overriding? I did a quick search and could see a few "override ToString" so wasn't sure what to do.

[imback82]

Could you elaborate on the use case?

[GoEddie]

In spark toString is overridden (not sure if it is always or just sometimes) so doing:

https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/Column.scala#L153

col("a").toString

returns a string "a"

if you do that on a Col in .NET you get:

Col("a").ToString()

returns a string "Microsoft.Spark.Sql.Column".

It feels like if we forwarded ToString and Equals then we will be more consistent with whatever spark does (or doesn't if it doesn't override them in some cases).

[imback82]

Got it. Yea, I think we can forward them to Scala side.

[GoEddie]

Thanks, Will do

[imback82]

I am curious what the notebook experience will be like since it depends on ToString. @suhsteve any concern?

[suhsteve]

There may be some unintended side effects by override Equals ? One example may be if a user is adding/checking these jvm objects against C# dictionary/hashset, any collisions would create jvm calls.

[suhsteve]

I am curious what the notebook experience will be like since it depends on ToString. @suhsteve any concern?

I don't see anything too concerning with toString with notebooks atm.

[imback82]

I am thinking we should support Equals and ToString for both C# side and JVM side (C# as the first citizen and a variation of Equals and ToString for the Scala). What do you guys think?

[GoEddie]

I’d say pass toString on, this is another example: #594

[imback82]

Yea, we can start with ToString. Should we pass thru for all the wrapper classes or the ones that override toString on the Scala side? (I see that we have few classes that override ToString to call the Scala side b/c it's overridden on the Scala side).