[jira] [Commented] (SPARK-8288) ScalaReflection should also try apply methods defined in companion objects when inferring schema from a Product type

["Drew Robb (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SPARK-8288?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16107798#comment-16107798 ] 

Drew Robb commented on SPARK-8288:
----------------------------------

I opened a PR for this issue, not sure why the bot didn't pick it up? https://github.com/apache/spark/pull/18766

> ScalaReflection should also try apply methods defined in companion objects when inferring schema from a Product type
> --------------------------------------------------------------------------------------------------------------------
>
>                 Key: SPARK-8288
>                 URL: https://issues.apache.org/jira/browse/SPARK-8288
>             Project: Spark
>          Issue Type: Improvement
>          Components: SQL
>    Affects Versions: 1.4.0
>            Reporter: Cheng Lian
>
> This ticket is derived from PARQUET-293 (which actually describes a Spark SQL issue).
> My comment on that issue quoted below:
> {quote}
> ...  The reason of this exception is that, the Scala code Scrooge generates is actually a trait extending {{Product}}:
> {code}
> trait Junk
>   extends ThriftStruct
>   with scala.Product2[Long, String]
>   with java.io.Serializable
> {code}
> while Spark expects a case class, something like:
> {code}
> case class Junk(junkID: Long, junkString: String)
> {code}
> The key difference here is that the latter case class version has a constructor whose arguments can be transformed into fields of the DataFrame schema.  The exception was thrown because Spark can't find such a constructor from trait {{Junk}}.
> {quote}
> We can make {{ScalaReflection}} try {{apply}} methods in companion objects, so that trait types generated by Scrooge can also be used for Spark SQL schema inference.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org