On 6/20/11 23:32 , Niclas Hedhman wrote:
Well, the Specification still needs to be 'translated' when there is
an underlying Query engine (most cases for performance). I am not
happy to see a trend that "always use underlying systems query
mechanism", since that binds very hard to a particular implementation
(trying to help people not shooting themselves in the foot). And
making 'programmable' native queries is a very dangerous tool which
people quickly go to... Just saying!

Here's the thing. We introduced NamedQueries as the way to do custom things when needed. The first thing I needed it for was Solr integration, where the query is simply a search string. And the NamedQueries was absolutely horrible to use for it and I had to "hack" a bit to get around it. So, I think it is better to allow for this to be easy, but make it clear that the operators are custom (by putting them in e.g. SolrExpressions.search(<string>)), rather than making it very hard to do.

/Rickard

_______________________________________________
qi4j-dev mailing list
[email protected]
http://lists.ops4j.org/mailman/listinfo/qi4j-dev

Reply via email to