New versions of Lucene Core and Solr are now available from the Apache Lucene project.
The Apache Lucene project is a collection of open-source search software consisting of Lucene Core, Solr, PyLucene and Open Relevance Project.
Lucene Core, formerly Lucene Java, is a Java-based indexing and search tool that also offers spellchecking, hit highlighting and advanced analysis/tokenization capabilities.
Solr is an enterprise search server with a number of APIs and search features including faceted and distributed search, along with admin and search interfaces. PyLucene is a Python port of the Lucene Core project.
The new version of Lucene (3.3) has better spellchecking, support for merging results from multiple shards, and an optimized implementation of KStem, a less aggressive stemmer for English. The spellchecker now has suggestions for misspelled words and offers auto-completion of words. It has three implementations: Jaspell, Ternary Trie, and Finite State.
A new option called NRTManager makes it easier to handle near-real-time search with multiple search threads, so allowing the application to control which indexing changes must be visible to which search requests.
Solr 3.3 has improvements including the ability to collapse groups and fields; an automaton-based suggest/autocomplete implementation that the release notes say offers an order of magnitude smaller RAM consumption. Solr now also defaults to a more efficient merge policy (TieredMergePolicy) that makes indexing more efficient.
For the full list of new features in Lucene and Solr and to download the software visit the Lucene Project site.