[DRILL-6217] NaN/Inf: NestedLoopJoin processes NaN values incorrectly

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.12.0
• Fix Version/s: 1.13.0
• Labels: ready-to-commit
• Assignee: Volodymyr Tkach
• Reporter: Volodymyr Tkach
• Reviewer: Arina Ielchiieva
• Created: 07/Mar/18 08:42
• Updated: 12/Mar/18 09:32
• Resolved: 10/Mar/18 15:41

Description

AFFECTED_FUNCTIONALITY: INNER JOIN (nestedloopjoin)

ISSUE_DESCRIPTION: according to nestedloopjoin query result NaN != NaN, however hashjoin / mergejoin behaves another way - NaN = NaN. As far as I understand, nestedloopjoin should behave like hashjoin / mergejoin. STEPS:

• Upload the attached file to Hadoop fs (ObjsX.json);
• Setup the following system settings:
  set planner.enable_nljoin_for_scalar_only = false
  set planner.enable_hashjoin = false
  set planner.enable_mergejoin = false
  set planner.enable_nestedloopjoin = true
• Run the following sql query

   select distinct t.name from dfs.tmp.`ObjsX.json` t inner join dfs.tmp.`ObjsX.json` tt on t.attr4 = tt.attr4 

  EXPECTED_RESULT: It was expected to get the following result:

  	 object1
  	 object2
  	 object3
  	 object4
  	

ACTUAL_RESULT: The actual result is:

	 object2
	 object3
	 object4
	

Please investigate and fix

Issue Links

• links to: GitHub Pull Request #1154