GRRDL notes

This is mostly a reminder to myself, but also for Sandro and any  
interested parties.

http://chatlogs.planetrdf.com/swig/2009-03-19#T17-18-49

This validates, I think, our use of a server side transform.

Also, it looks like we can output NTriples instead of RDF/XML. The  
spec allows it (scrollup in that chat transcript) and major engines  
(like Jena) accept it.

I would prefer we used XSLT 2.0 so we can at least make it somewhat  
typesafe.

 From a security perspective, it seems that Jena puts up a warning at  
least the first time you use GRDDL, but it's unclear if it does it  
every time it downloads a new transform. I don't know if it caches,  
so the effect on W3C traffic is still unknown. I don't know anything  
momre about signing or checksumming the XSLT, so I think it still is  
a fairly large security risk.

My (subjective) impression from the #swig chat is that there are  
active portions of the GRDDL community who are less...militant about  
the form of the transform and its mode of delivery. So, we should  
consider whether the vocal pro-specific-transform-delivered- 
specifically are, in fact, representative.

Obviously, this is subject to confirmation bias on my part. But given  
that several (though not *all*, by any means!) of the vehement pro- 
GRDDL people are also, independently anti-OWL/XML and not obvious  
users of it, I do believe that we do not need to go overboard wrt  
catering to their demands.

This does not mean I am, by any means, backing out of the tripartite  
compromise. But it does mean that my motivation for the compromise is  
weakened compared to the worries I have about GRDDL (esp. at this  
point, the security issues).

Cheers,
Bijan.

Received on Friday, 20 March 2009 12:13:13 UTC