Re: Java API for GRDDL processing and namespace question about GRDDL

On Nov 9, 2006, at 4:49 PM, Gokhan Soydan wrote:
> Dan,
>
> I downloaded and ran your glean.py program, and kind of found the 
> answer to
> my question about the result of loops. I see that the GRDDL result of 
> the
> original input document will have only the merge of the results of its 
> own
> dataview:transformations, namespacetransformations of its "immediate"
> namespaces and profiletransformations of its "immediate" profiles. So, 
> since
> the GRDDL result is just the transformations of the original document 
> and
> transformations from its "immediate" namespaces or profiles, which are
> listed in the original document, then why would there be a need to go
> through loops to get to other namespaces of namespaces of namespaces or
> profiles and so on?

Because to find the namespaceTransformations of the original
document, you should/may look at all the GRDDL results of
the namespace and profile document, not just their contents.

"A namespace transformation link may be discoverable by transforming 
the namespace document itself. Note that this means that namespace 
documents need not be written in RDF/XML directly."


> Also, wouldn't we have a problem if:
>
> ?D has namespaces ?NS and ?NSS

That's impossible; there's at most one namespace associated
with the root of D. It could happen with profiles, though...

> ?NS has namespace ?NSS
>
> The program first visits ?NS from ?D, and puts ?NS on the already list.
> Then, immediately, program visits ?NSS from ?NS, and puts ?NSS on the
> already list. When the program returns back to ?D, and continues to get
> namespaces, this time it gets ?NSS, but ?NSS is in the already list 
> from
> earlier, so it doesn't visit ?NSS from ?D. This could be a problem if 
> ?NSS
> has namespacetransformation for ?D.

I don't think so; the already list is a parameter on the call stack, 
which
gets popped when returning back to ?D.

It's probably worth testing, though.
I added it to a TODO list in ...
  http://www.w3.org/2001/sw/grddl-wg/td/testlist2
  v 1.2 2006/11/10 04:09:48


> Thanks,

Likewise, thanks for the careful review.

p.s. somebody from topquadrant (Holger?) gave me a demo of TopBraid's
GRDDL support at ISWC this week. It worked just fine for the case we 
tried.
Keep up the great work!

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Friday, 10 November 2006 04:12:30 UTC