From: Ron V. d. B. <ron...@ka...> - 2010-10-26 08:00:15
|
Hi Anne, On 25/10/2010 18:08, Anne wrote: > Did you happen to have read my comment on your blog > (http://rvdb.wordpress.com/2010/10/06/mimicking-faceted-searching-in-exist/), > which I posted just before I read this? It might be a related issue? > I don't know they're related, but don't think so. > For completenes, my comment boils down to this: > > I was trying to reconstruct the faceted search examples on the blog, > to compare the result to my own implementation (using a new type of > index) of faceted search in eXist. > It turns out that if I follow the example literally, I get no results > for the approach listed under “A simple facet with eXist-specific > functions”. I do get hits, but no statistics about speakers. Is it me, > or is something missing? Apparently, there's another bug in eXist trunk, where util:index-key() only seems to work on path-based range indexes. It has been reported twice on the ML: -<http://markmail.org/message/aebx4mhfu5dymilw> -<http://markmail.org/message/lqtzzb6ftqcrzvlt> For the time being, you can work around the issue with my query by a) either experimenting with eXist-1.4.x (which doesn't have the problem), or b) redefining the range indexes with a path expression: <!-- range indexes --> <create qname="SPEAKER" type="xs:string"/> <create qname="TITLE" type="xs:string"/> After reindexing, the statistics are returned properly. Perhaps I'd better change this index definition in my blog post as well. Kind regards and good luck with your work (I'm very curious!), Ron |