« So, that took a while to figure out | Main | Operational mindset »

Off by one

February 28, 2007 | co.comments icon co.mments

"Doesn't matter where it came from - one source might be the result of a SPARQL CONSTRUCT query against a bunch of scuttered material."

The point isn't where the data came, it's that the data has been organized and marshaled in advance into a domain model before being passed to the rendering layer. It has to be, or the basic publishing exercise becomes too complicated - in Danny's example everything becomes a for loop filtering over value streams instead of simple field gets on an object. RDF backed data needs to be fitted into a domain before it plays nice with the web stack - anything else is a rip and place of at least two layers. I've been saying this for a few years now. Trying his example against arbitrary graphs will get a feel for the richness involved. The problem is some people will point out that if the data is organized into domain models, handling/storing arbitrary RDF isn't necessary. IOW, if the vcard example was actually realistic, RDF isn't warranted.


February 28, 2007 10:30 PM

Comments

Post a comment

(you may use HTML tags for style)




Remember Me?

Trackback Pings

TrackBack URL for this entry:
http://www.dehora.net/mt/mt-tb.cgi/2048