[HIVE-10215] Large IN() clauses: deep hashCode performance during optimizer pass

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 1.2.0
• Fix Version/s: 1.2.0
• Component/s: Logical Optimizer
• Release Note: Use object identity to prevent recursion instead of equality in the optimizer visitor pattern
• Assignee: Gopal Vijayaraghavan
• Reporter: Gopal Vijayaraghavan
• Created: 04/Apr/15 07:23
• Updated: 25/Aug/15 03:53
• Resolved: 11/Apr/15 08:20

Description

The logical optimizer uses several maps and sets, which are exceeding expensive for large IN() clauses due to the fact that several part of the queries walk over the lists without short-circuiting during hashCode(), while equals() is faster due to short-circuiting via less expensive operators.