[SPARK-18674][SQL][Follow-Up] improve the error message of using join

[gatorsmile]

What changes were proposed in this pull request?

Added a test case for using joins with nested fields.

How was this patch tested?

N/A

[gatorsmile]

cc @cloud-fan@hvanhovell

[SparkQA]

Test build #69513 has finished for PR 16110 at commit e754ef5.

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

[gatorsmile]

retest this please

[cloud-fan]

shall we put the test in ResolveNaturalJoinSuite and follow the code style there?

[rxin]

OK if you are improving this, can you fix the grammar? "USING column col1.field1 can not be resolved with the left join side, the left output is: [col1]" technically is grammatically incorrect.

[gatorsmile]

Sure, will do it. Thanks!

[SparkQA]

Test build #69522 has finished for PR 16110 at commit e754ef5.

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

[gatorsmile]

Just tried MySQL. The message is not good to me:

mysql> select * from t1 inner join t2 using (col2);
ERROR 1054 (42S22): Unknown column 'col2' in 'from clause'

Will try to improve it by:

Using column `col1.field1` cannot be resolved on the left side of the join. The left-side columns: [col1]

[rxin]

That SGTM

[SparkQA]

Test build #69548 has started for PR 16110 at commit 4d43bca.

[rxin]

LGTM pending tests.

[gatorsmile]

retest this please

[SparkQA]

Test build #69554 has finished for PR 16110 at commit 4d43bca.

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

[cloud-fan]

thanks, merging to master/2.1!