Skip to content

Apache Lucene open-source search software

Move the #16682 entry to 10.6.0, where it ships first

Jim Ferenczi3be9cd6e6629
diff --git a/lucene/CHANGES.txt b/lucene/CHANGES.txt
index b767db4e58c..4d0c74d2cbc 100644
--- a/lucene/CHANGES.txt
+++ b/lucene/CHANGES.txt
@@ -192,12 +192,6 @@ Improvements
Optimizations
---------------------
-* GITHUB#16682: Carry file open hints on merge and flush contexts, which used to drop them. An HNSW
- format now carries the random access hint for the vectors it reads, rather than the flat format
- holding them claiming it whatever reads them. Add NoReuseHint, set by stored fields, term vectors
- and the raw vectors kept only to rescore, so an application can tell those apart from the files it
- wants to keep in memory. (Jim Ferenczi)
-
* GITHUB#16431: Add support for Block-Max WAND dynamic score skipping in FunctionScoreQuery
and DoubleValuesSource using DocValuesSkipper. (Rajat Jain)
@@ -465,6 +459,12 @@ Improvements
Optimizations
---------------------
+* GITHUB#16682: Carry file open hints on merge and flush contexts, which used to drop them. An HNSW
+ format now carries the random access hint for the vectors it reads, rather than the flat format
+ holding them claiming it whatever reads them. Add NoReuseHint, set by stored fields, term vectors
+ and the raw vectors kept only to rescore, so an application can tell those apart from the files it
+ wants to keep in memory. (Jim Ferenczi)
+
* GITHUB#16593: DenseLiveDocs and SparseLiveDocs implement Bits#applyMask in bulk, restoring the
word-at-a-time masking that live docs had before GITHUB#13084 exposed them as LiveDocs rather than
as a FixedBitSet, so AcceptDocs, the bulk scorers and CheckIndex no longer test one bit per