On Fri, Feb 7, 2014 at 3:32 PM, anand chandak <anand.chan...@oracle.com>wrote:
> Thanks Mikhail, curious why was scoring left out of solr ? Have no idea. I've never been involved in query-time join. > And if there's any plan to port it ? > I suppose there is no plan until someone raise a jira. > Also, if you can please elaborate on the segmented vs toplevel > http://vimeo.com/44113003 "Is Your Index Reader Really Atomic or Maybe Slow?" by Uwe Schindler > Thanks, > > Anand > > > > On 2/7/2014 4:53 PM, Mikhail Khludnev wrote: > >> Basically, i am trying to understand where and how solr joins differ from >>>> lucene joins. Any pointers, much appreciated ? >>>> >>>> >>>> Hello Anand, >> >> I'm keen for index-time joins (aka block joins), thus I've never looked >> into query-time ones. >> I even didn't ever know that there are two different query-time joins. >> This >> diverging might caused the fabulous drama: segmented vs top-level >> Ok. It seems like, Solr's query time join never scores >> https://github.com/apache/lucene-solr/blob/trunk/solr/ >> core/src/java/org/apache/solr/search/JoinQParserPlugin.java#L535 >> But Lucene's JoinUtils does >> https://github.com/apache/lucene-solr/blob/trunk/lucene/ >> join/src/java/org/apache/lucene/search/join/JoinUtil.java?source=cc#L78 >> Also Solr's join joins across different cores. >> >> > -- Sincerely yours Mikhail Khludnev Principal Engineer, Grid Dynamics <http://www.griddynamics.com> <mkhlud...@griddynamics.com>
