Re: RDFON: a new RDF serialization

>  I'm not so sure I agree with the syntax-is-the-problem tradition. Surely 
>  syntax is one issue, but I tend to think it was just as much lack of a) 
>  well-developed tool infrastructure (since you should be working with models, 
>  not syntax), and b) a standard query language.

the agile / ruby / python / microformats / web 2.0 developer communities are born from 'languages simple enough that you can write it in notepad' (well, they use textmate). this precludes requiring an IDE just to edit stuff, or XML libraries to turn unreadable gunk into in-memory models.

>  Both of those are being resolved.

can you explain why one query language is a good thing. especially when it introduces another (!) syntax to the equation, and is heavily dependent on argument order for performance. surely exposing basic pattern matching primitives via a REST interface, and letting the user specify queries in XML or JSON does a better job of letting a semantic web organically grow while reusing existing parts, than 'prescribing' a new solution that also requires a new tool chain for parsing/serializing/editing... 

>  In my experience, people who complain about the RDF/XML syntax complain 
>  about XML too. 

i dislike XML quite a bit, but i dont have a problem with RDF/XML. if redland didnt exist, i wouldn't support it for import and export though..


>  Um ... sure, but what's wrong with N3 or turtle that requires yet another 
>  slightly different syntax?
>  Isn't there a JS-based RDF parser that can handle N3?

there are a few. but none are as fast as JSON, since that uses a fast VM level eval(), instead of JS level string parsing regex stuff..

so JSON gets you the readablity of Turtle without the speed hit of even N3..

Received on Thursday, 26 July 2007 22:32:30 UTC