Apache Lucene 6: What's coming next? #bbuzz
summary
Lucene 6 introduces points (dimensional values) as a new, codec-backed data type replacing NumericRangeQuery for faster 1–8D queries (IPv6/128‑bit ranges supported), reduces index size and heap use, and speeds indexing/search; the Filter class is removed in favor of non‑scoring clauses inside BooleanQuery, two‑phase iterators enable cheap/expensive checks, and docValues replace the field cache. Runtime requires Java 8 minimum requirement, scoring switches to Okapi BM25 default, and Solr 6 adds an SQL dialect plus a JDBC driver and async cross‑datacenter replication.
AI-written summary · about
chapters (12)
- 0:00Introduction and background
- 1:53Lucene 5 data safety features
- 7:03Lucene 5 codec API overhaul
- 13:09Field cache removal and doc values
- 16:42Deprecation of filter class
- 19:15Two-phase iterators in queries
- 22:35Lucene 6 point values introduction
- 32:29Java 8 requirement and profiles
- 33:53BM25 scoring algorithm in Lucene 6
- 35:40Query immutability and changes
- 36:44Solr 6 features overview
- 39:36Q&A
related talks
22mWhat's coming next with Apache Lucene?
41mApache Lucene and Java 9+ #bbuzz
41mApache Lucene 5 - New Features & Improvements for Apache Solr
37mShipping Lucene 10.0, 25 years in the making
43mWhat's new in Apache Solr 9.0
37mFast Decompression Lucene Codec #bbuzz
39mAccelerating Aggregations with Lucene-Aware Execution in OpenSearch
51mHibernate Search talk