[LUCENE-6548] Optimize BlockTree's Terms.intersect a bit for "very finite" automata

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 5.3, 6.0
• Lucene Fields: New
• Assignee: Michael McCandless
• Reporter: Michael McCandless
• Created: 11/Jun/15 09:47
• Updated: 28/Aug/22 14:34
• Resolved: 23/Jun/15 20:15

Description

I've been digging into why BlockTree's Terms.intersect impl is slower for a "very finite" union-of-terms test over random IDs (LUCENE-3893) and I found a few performance improvements that at least for that one use case gets a ~11% speedup.