Re: Review of testlist1#rdfa1 (and testlist2#multiprofile)

On Sun, 2006-12-17 at 12:08 -0500, Ben Adida wrote:
> Dan Connolly wrote:
> > I think that's overstating the situation a bit. I don't think this WG
> > made any decisions about hGRDDL. (All our decisions are
> > in our meeting minutes; if I'm forgetting one, a pointer will
> > refute my claim conclusively.)
> 
> I may be confused about what was discussed in offline conversations and
> what happened on the mailing list. I should have double-checked. Indeed,
> it doesn't look like there is a decision on hGRDDL in the GRDDL public
> space.

Right.

>  Maybe we can recreate hGRDDL effectively from existing
> features... that would be fantastic.

If not, we'll learn something by trying...

> > Just use the normal GRDDL markup, just like the Atom/turtle test
> > case and our decision on issue-output-formats shows.
> > An RDFa document is a serialization of an RDF graph just like a
> > turtle document is, no?
> 
> Maybe I'm confused by Ian's recent message that says that he's not sure
> how the proper URI would be obtained by an RDFa parser [1]. If it's
> possible, then that's great.
> 
> I'm looking at the Turtle example. I see the grddl:transformation triple
> specifies the XSLT. The question is, can different transformations be
> specified in the profile document, which the parser might be able to
> select based on desired output format? From my read of the mailing list,
> I think not.

Right; but do you need such a thing? The author can just include
multiple transformations. The consumer can just try them all.

I'm getting a little confused about the details. Maybe you can
sketch an example?

> > Is that stuff really not specified to your satisfaction?
> 
> I don't see specifications of what happens when there are multiple
> PROFILEs in the HEAD.

Hmm... didn't I have a figure about that? No... that was
multiple <link rel="transformation">s. You're right; the spec
isn't very clear on this...

"if an information resource ?D has an XHTML representation whose profile
attribute refers to ?PROFILE ... "

 -- http://www.w3.org/2004/01/rdxh/spec#profile-bind

I added a @@TODO to fix that in 1.174.

Let's see... Dom made a test case...
  xhtmlWithMoreThanOneProfile

Ah... but that's a test for robustness...
<head profile='http://www.example.org/NotAGrddlProfile
http://www.w3.org/2003/g/data-view
http://www.example.org/AnotherUndereferencableProfile'>

OK, I just cobbled together a test that shows that with 2 profiles,
you find profileTransformations for each, run them, and merge
the results:
  http://www.w3.org/2001/sw/grddl-wg/td/testlist2#multiprofile
  http://www.w3.org/2001/sw/grddl-wg/td/multiprofile.html
  http://www.w3.org/2001/sw/grddl-wg/td/multiprofile-output.rdf



>  I see what happens when there are multiple
> <LINK>s. IIRC, there was a discussion about how each profile would be
> processed separately, not pipelined or combined, but I can't find the
> relevant thread.

There's no serial interaction, but the results are combined,
i.e. merged.

"If F and G are GRDDL results of IR, then the merge [RDF-MT] of F and G
is also a GRDDL result of IR"

I'm not sure where pipelining would come into play.
If you have a transformation that produces RDFa, it's just
like a transformation that produces turtle: as long
as the consuming agent understands how to get an RDF graph
from the result of the transformation, it works; i.e. if
it has a built-in RDFa parser, just like it might
have a built-in turtle parser. More agents
will know how to consume RDF/XML than other formats, so
there's a deployment risk, but the design is reasonably clear,
I think.

> Note that "to my satisfaction" is the wrong qualifier. There are some
> issues that I care about that the WG has not decided to make part of its
> scope, and that's fine, GRDDL can't be everything to everyone. That
> said, maybe everything I care about happens to be solved by the existing
> spec, which would be great. I would be extremely happy if I had misread
> the spec in that respect :)
> 
> -Ben
> 
> [1] http://lists.w3.org/Archives/Public/public-grddl-wg/2006Dec/0043
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Monday, 18 December 2006 21:08:52 UTC