[SPARK-9295] Analysis should detect sorting on unsupported column types

[JoshRosen]

This patch extends CheckAnalysis to throw errors for queries that try to sort on unsupported column types, such as ArrayType.

[SparkQA]

Test build #38306 has finished for PR 7633 at commit 2f1b802.

• This patch fails Spark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[JoshRosen]

Ah, it looks like this broke sorting by a null literal, which is valid SQL and which should be supported. I think that we can fix this by making NullType into an AtomicType (I think we've already done this in Python).

[JoshRosen]

Actually, I think I'll just explicitly allow NullType.

[JoshRosen]

Jenkins, retest this please.

[SparkQA]

Test build #38328 has finished for PR 7633 at commit bfe1451.

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

[rxin]

why not just embed this in foreach?

[JoshRosen]

This matches the style of Aggregate above.

[rxin]

I think aggregate was only doing it because it was recursive. Not that big a of deal.

[JoshRosen]

Meh, I can update it.

[SparkQA]

Test build #38373 has finished for PR 7633 at commit 23b2fbf.

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

[JoshRosen]

I'm going to merge this into master. Thanks!