Re: Why is RDF such a tough sell?

I'd just like to throw in a few thoughts from a real life example
and why we didn't go with RDF.  Trust me, I wanted to.


We're developing a distributed
(as in many little linux devices on many different networks, and they 
all have to communicate with a central system) application and needed to 
express objects back and forth.  Our central system is written in Java, 
and our devices run linux but only have libc on it, so we're stuck in C 
(due to footprint constraints).  We have a very close deadline, so "can 
I do this now?" was the question we were asking.

I proposed looking at RDF, because I want to transfer objects across and 
I want a very loose coupling between these devices (which will change 
and will become smarter over time) and central.  With DAML+OIL I can 
cope with changes in the ontology.  In theory, it all worked out quite 
nicely, and the team loved it.

We had to scrap RDF because we couldn't find a working rdf toolkit 
written in C that could open a URI over the network, get an RDF stream 
(as XML or triples, we don't care), parse it, store it and later on, 
serialize it back (to XML or triples), and send it back over the wire to 
a URI (we are very REST here).

In short, my point is, we all "got it" about RDF, but couldn't for 
practical reasons.  There just aren't RDF toolkits for C out there 
(without hacking together a URI/network library into an RDF parser into 
a good storage with querying, and back out again).  I believe RDF is 
tough sell /because you can't do it yet/.  Too much glue right now for a 
real life project.  Everyone here wanted to use RDF and DAML+OIL for 
transfering information but couldn't without a LOT more code, and we're 
not in the business of writing RDF frameworks.  (although that would be 
pretty fun)

I hope someone proves me wrong and says "Duh, here is that library you 
are looking for" so next time we can do it right. :)

Not to be so gloomy, but a real RDF solution isn't practical now for 
fast moving projects.

Seth

Received on Tuesday, 25 June 2002 17:32:50 UTC