[SPARK-17053][SQL] Support hive.exec.drop.ignorenonexistent

[dongjoon-hyun]

What changes were proposed in this pull request?

If hive.exec.drop.ignorenonexistent is set true, DROP statement is assumed to have IF EXISTS clause.

Reported Error Scenario

scala> sql("set hive.exec.drop.ignorenonexistent=true")
res1: org.apache.spark.sql.DataFrame = [key: string, value: string]

scala> sql("drop table a")
org.apache.spark.sql.AnalysisException: Table to drop '`a`' does not exist;

Refer https://cwiki.apache.org/confluence/display/Hive/LanguageManual+DDL .

How was this patch tested?

Pass the Jenkins test including a new test case.

[jaceklaskowski]

The JIRA issue (https://issues.apache.org/jira/browse/SPARK-17053) is closed as Won't Fix. Should the PR be closed too?

[dongjoon-hyun]

Nope, @jaceklaskowski . The discussion is not ended. The need of user is just supporting this option to support hive compatibility. I reopened the issue since he does not give any reason clearly.

[SparkQA]

Test build #63752 has finished for PR 14636 at commit 0d25b25.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[dongjoon-hyun]

Hi, @rxin .
Could you review this PR about hive.exec.drop.ignorenonexistent?

[dongjoon-hyun]

I'm close this issue since Spark does not want to support this option. The issue was finalized as WON'T FIX. See the Jira issue.