[DRILL-1169] Add support for UNION (distinct type)

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.1.0
• Component/s: Query Planning & Optimization
• Labels: doc-impacting

Environment

Running Drill on MacOSX standalone.
#Sun Jul 20 20:12:53 PDT 2014
git.commit.id.abbrev=e5c2da0

• Assignee: Sean Hsuan-Yi Chu
• Reporter: Bob Rumsby
• Created: 22/Jul/14 20:42
• Updated: 15/Jun/15 14:22
• Resolved: 15/Jun/15 14:22

Description

A query that selects a compatible column from the same JSON file on either side of the UNION operator fails. The same query works if UNION ALL is used.

Example of Failure:
0: jdbc:drill:zk=local> select id from dfs.`/Users/brumsby/drill/donuts.json` union select id from dfs.`/Users/brumsby/drill/donuts.json`;
Query failed: org.apache.drill.exec.rpc.RpcException: Remote failure while running query.[error_id: "dfe76b6b-7eb0-49ee-8084-1e29ea995be0"
endpoint {
address: "10.250.0.28"
user_port: 31010
control_port: 31011
data_port: 31012
}
error_type: 0
message: "Failure while parsing sql. < CannotPlanException:[ Node [rel#880:Subset#6.PHYSICAL.SINGLETON([]).[]] could not be implemented; planner state:
...

Same query with UNION ALL (which works):

0: jdbc:drill:zk=local> select id from dfs.`/Users/brumsby/drill/donuts.json` union all select id from dfs.`/Users/brumsby/drill/donuts.json`;
------------

id

------------

0001

0001

------------
2 rows selected (0.111 seconds)