[SPARK-12719][HOTFIX] Fix compilation against Scala 2.10

[yy2016]

What changes were proposed in this pull request?

Compilation against Scala 2.10 fails with:

[error] [warn] /home/jenkins/workspace/spark-master-compile-sbt-scala-2.10/sql/hive/src/main/scala/org/apache/spark/sql/hive/SQLBuilder.scala:483: Cannot check match for         unreachability.
[error] (The analysis required more space than allowed. Please try with scalac -Dscalac.patmat.analysisBudget=512 or -Dscalac.patmat.analysisBudget=off.)
[error] [warn]     private def addSubqueryIfNeeded(plan: LogicalPlan): LogicalPlan = plan match {

How was this patch tested?

Compilation against Scala 2.10

[AmplabJenkins]

Can one of the admins verify this patch?

[gatorsmile]

You are using a wrong JIRA number.

[JoshRosen]

Why not just increase the pattern match analysis budget?

[yy2016]

This was the first build which failed:
https://amplab.cs.berkeley.edu/jenkins/view/Spark%20QA%20Compile/job/SPARK-master-COMPILE-sbt-SCALA-2.10/500/

There was only one checkin:
[SPARK-12719][SQL] SQL generation support for Generate

[yy2016]

bq. Why not just increase the pattern match analysis budget

There is a chance the new budget gets exceeded in the future when more cases are added.

[yhuai]

Thanks. I am merging this to master.

[yhuai]

Just reverted it because it breaks the build.

[yy2016]

I wonder why OneRowRelation isn't covered by the following import ?

import org.apache.spark.sql.catalyst.plans.logical._

[liancheng]

@yy2016 OneRowRelation is an object rather than a class, and should be matched using case OneRowRelation =>.

[yy2016]

@liancheng
What's your preference ?
@JoshRosen suggested defining Dscalac.patmat.analysisBudget

[liancheng]

I agree with @JoshRosen. Was just answering your question above :)