You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@lucene.apache.org by Apache Wiki <wi...@apache.org> on 2017/01/17 19:29:55 UTC
[Lucene-java Wiki] Update of "ReleaseNote64" by MikeMcCandless
Dear Wiki user,
You have subscribed to a wiki page or wiki category on "Lucene-java Wiki" for change notification.
The "ReleaseNote64" page has been changed by MikeMcCandless:
https://wiki.apache.org/lucene-java/ReleaseNote64?action=diff&rev1=10&rev2=11
Lucene 6.4.0 Release Highlights:
New features
+ * Lucene's best efforts to un-map memory mapped files with "!MMapDirectory" now work with the latest Java9 early access builds
- * A new similarity "!BooleanSimilarity that gives terms a score that is equal to their query boost
+ * A new similarity "!BooleanSimilarity" that gives terms a score that is equal to their query boost
+ * The axiomatic family of similarities (6 in total) based on https://www.eecis.udel.edu/~hfang/pubs/sigir05-axiom.pdf
* A new analyzer "!SynonymGraphFilter" that outputs a correct graph structure for multi-token synonyms at query time
* Graph token streams produced by the "!SynonymGraphFilter" are now handled accurately by query parsers
* A new collector "!DocValuesStatsCollector" gives the ability to compute statistics on DocValues field
* It is now possible to filter "!SortedDocValues" and "!SortedSetDocValues" terms enum with a compiled automaton
* The "!UnifiedHighlighter" can now highlight fields with queries that don't necessarily refer to that field
* DrillSideways can now run queries concurrently
+ * Index sorting now supports sorting on multi-valued fields using MIN, MAX, etc. selectors
Optimizations
* Points do not store the implicit split dimension in the 1-dimension case. This saves between 6% memory for the largest types such an !InetAddressPoint to 33% for the smaller types such as !HalfFloatPoint.
* The BKD in-memory index for dimensional points now uses a compressed format, using substantially less RAM in some cases
* The BKD writing now buffers each leaf block in heap before writing to disk, giving a small speedup in points-heavy use cases
+ * "!TermAutomatonQuery" now rewrites to more efficient queries when possible
+ * Index sorting now sorts on flush instead of first merge, resulting in sizable speedups: see annotation V in https://home.apache.org/~mikemccand/lucenebench/sparseResults.html#index_throughput
Along with numerous bug fixes.