Hi, I am working on a migration from verity k2 to solr.
At this point I have a parser for the Verity Query Language (our used subset) which generates a syntax tree. I transfer this in a couple of filters and one query. This fragmentation is the reason, why I can not use my parser inside solr (via QparserPlugin: http://wiki.apache.org/solr/SolrPlugins#QParserPlugin ). Because I have a syntax tree I like to use the QueryParser Lucene contrib module. Other reason is, that we need our own PhraseQuery, so the normal Solr Query-Syntax will not work (not even the nested queries http://www.lucidimagination.com/blog/2009/03/31/nested-queries-in-solr/ http://www.lucidimagination.com/blog/2009/02/22/exploring-query-parsers/ ) So lets say that I already have a org.apache.lucene.queryParser.core.nodes.QueryNode What is the proper way to send this to solr? 1. serialize to XML and use xml-query-parser for deserialization (but support in Solr is not stable: https://issues.apache.org/jira/browse/SOLR-839 ) 2. serialize and deserialize with XStream 3. serialize and deserialize with NamedList (like SolrJ does this in the other direction) 4. other suggestions? If 1.: Does anyone use xml-query-parser with heavy loads? If 2.: Does anyone know why QueryNodeImpl.java lost its serialVersionUID from 3.x to 4.x? Will it not longer implement Serializable? If 3.: Does anyone used NamedList to sent informations to solr? Does anyone used NamedList to represent a QueryNode (or syntax tree)? Best regards Karsten P.S. An small example why QueryParser is great: http://sujitpal.blogspot.com/2011/03/using-lucenes-new-queryparser-framework.html
