This release of Lux is integrated with Solr/Lucene 4.2
Added low-level support for binary documents; you can now store images, xquery, and other non-XML files in a Lux index.
Fixed a bug where a document could be stored twice at the same uri, with different content.
Index-assisted sorting by numeric fields
Enabled configurable analysis chain so users can control the kind of text treatment performed by the XML indexing analyzers. Specifically, XmlTokenStreamBase now wraps an externally-supplied Analyzer. We’ll need to follow up by exposing higher-level constructs to make this more usable, eg: expose via configuration in schema.xml. This may actually work already and just require documentation?
XPathField now uses the analysis chain configured in schema.xml
Optimized deep pagination: we now skip over (not load into memory and parse) documents we can prove to be unused.
Fixed an over-optimization bug when we optimized count (exists, not) and its argument combined a search call and an additional optimizable expression (like a path).
Added “querybox” - a query sandbox - to the demo.
Fixed bug where we dropped the return types of user-defined functions, including their cardinalities
Fixed a variable-shadowing bug
Tested with a multi-core Solr install; seems to work OK.
Use an LRUCache in CachingDocReader to limit memory usage