On 2010-04-28 17.46, Stanislav Muhametsin wrote:
Exactly. That's why I was thinking of my current work on RDBMS being indexing-only storage, with the primary storage being something else. I still think that actual querying can be done better using RDBMS, partly because most, even free, implementations of RDBMS are so highly optimized, unlike what I've seen in RDF world. Of course that is partly because RDBMS world is older than RDF.
Ok, cool!
Speaking of that - how large you would estimate your entitites to be QName-wise? Ie, including value composites, how many QNames are there in your entities in your application on average? "You" here means anyone who is reading this and has some kind of Qi4j-app going on.
A typical nr for me is about 10-15 for an average entity. Each mixin will have about 1-3 fields each.
I guess the point is to add RDBMS support to Qi4j to help it spread wider. That's how I remember it from some old mail I read which made me start doing work on this. I think it was near the time when Qi4j 1.0 was released. Plus - this has been extremely interesting learning process for me personally. :)
Ok, well, if you're going the table=mixin route, then maybe it could actually work out ok. At least both querying and storing *should* work!
It'll be interesting to see it in action! /Rickard _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
