Github user jpountz commented on a diff in the pull request:
https://github.com/apache/lucene-solr/pull/525#discussion_r243647937
--- Diff: lucene/core/src/test/org/apache/lucene/index/TestDocValues.java
---
@@ -123,8 +124,72 @@ public void testNumericField() throws Exception {
iw.close();
dir.close();
}
-
- /**
+
+ // The LUCENE-8585 jump-tables enables O(1) skipping of IndexedDISI
blocks, DENSE block lookup
+ // and numeric multi blocks. This test focuses on testing these jumps.
+ @Slow
+ public void testNumericFieldJumpTables() throws Exception {
+ Directory dir = newDirectory();
+ IndexWriter iw = new IndexWriter(dir, newIndexWriterConfig(null));
+ final int maxDoc = atLeast(3*65536); // Must be above 3*65536 to
trigger IndexedDISI block skips
--- End diff --
Why do we have this test here as opposed to TestLucene80DocValuesFormat?
Given that you are indexing so many documents and don't care about testing
indexing as much as you care about checking iterators, you should make sure
that the index writer config isn't too crazy and enforce sane values of
maxBufferedDocs, RAMBufferSizeMB, and a sane merge policy (eg. not the
alcoholic one). See for instance
TestLucene70DocValuesFormat#doTestSortedNumericBlocksOfVariousBitsPerValue.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
