[SPARK-5187][SQL] Fix caching of tables with HiveUDFs in the WHERE clause

[marmbrus]

No description provided.

[SparkQA]

Test build #25346 has started for PR 3987 at commit 8bca2fa.

• This patch merges cleanly.

[cfregly]

lgtm. as we just discussed, this is the same code path as SchemaRDD.cache(), so no need for additional tests.

[sarutak]

nit: should functionClassName be still var?

[marmbrus]

Yeah, its mutated below by our custom deserialization.
On Jan 9, 2015 4:31 PM, "Kousuke Saruta" [email protected] wrote:

In sql/hive/v0.13.1/src/main/scala/org/apache/spark/sql/hive/Shim13.scala
#3987 (diff):

@@ -53,7 +53,7 @@ import scala.language.implicitConversions
*

• @param functionClassName UDF class name
  */
  -class HiveFunctionWrapper(var functionClassName: String) extends java.io.Externalizable {
  +case class HiveFunctionWrapper(var functionClassName: String) extends java.io.Externalizable {

nit: should functionClassName be still var?

—
Reply to this email directly or view it on GitHub
https://github.com/apache/spark/pull/3987/files#r22753114.

[sarutak]

Ah, I see. It's externalaizable.

[SparkQA]

Test build #25346 has finished for PR 3987 at commit 8bca2fa.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • case class HiveFunctionWrapper(functionClassName: String) extends java.io.Serializable
  • case class HiveFunctionWrapper(var functionClassName: String) extends java.io.Externalizable

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/25346/
Test PASSed.

[marmbrus]

Merged to master and branch-1.2