Re: hope to hear from you today about the GRDDL spec

Dan,

Overall, I think this document is clear and complete. I have one
conceptual point to raise regarding sequential transformations, and a
bunch of small/medium bugs.

Given my tardiness in providing comments, I'm happy if these are raised
as issues TBD rather than fixed immediately. Most of the bugs should be
easy to address, though.

I highlighted two bugs with "***", because, unless I'm misreading the
formal statements, there are two significant bugs that may confuse the
reader when it comes to namespaceTransformation and
profileTransformation. These seem important to fix.

Thumbs up on publication.

-Ben

==============================

CONCEPTS
- Sequential Transformations. The idea behind the GRDDL output as
XHTML+RDFa (as defined in the use cases) is that it would be possible to
have the following sequential transformations:

XHTML+microformat --> XHTML+RDFa
to maintain the rendered text + metadata.

then
XHTML+RDFa --> RDF/XML
if someone wants the pure triples.

Now, since XHTML+RDFa will (eventually) be a valid RDF serialization,
maybe no one cares about the second transformation to RDF/XML. But I can
see how programs that are not yet RDFa-aware might want that second
transformation.

Right now, the spec doesn't talk about chaining transformations, yet the
issue will come up even if we choose not to address it. For example, I
can see one GRDDL transformation producing XML (possibly XHTML) that
itself contains a profile or GRDDL-transform indication (maybe in its
namespace document), that should then trigger a second GRDDL transformation.

We should explore this issue a bit at some point.

BUGS
- should we really use the RDF-in-HTML mailing list for public comments
on GRDDL? That doesn't seem right to me (We're always happy to get more
involvement from this group in the RDFa TF, but this doesn't seem like
the right way to do it :)

- "RDFa", not "RDF-a".

- 1. > Resource Descriptions
'One way to represent the fact "The Stand's author is Stephen King" is
RDF would be...'
--> '_in_ RDF would be...'

or, alternatively,
'...is an RDF triple whose subject is..."

- 1. > Faithful Renditions
"Specifying a GRDDL transformation is a statement by the author..."
I see what this means, but the formulation seems a bit awkward.
Suggestion:
"By specifying a GRDDL transformation, the author of a document states
that..."

Same suggestion for the second paragraph of that subsection "Likewise,
specifying a GRDDL namespace ..."

- 1. > GRDDL Primer
This subsection seems a bit terse.

- 2. > second bullet point in second paragraph
Inconsistency: use of ?T1 and ?T to reference the same transformation.

- 3. > first paragraph
"from a namespace URI _may is_ a namespace document"
probably "may be" or "is."


second paragraph
"To associate a _grddl_ transformation"
--> "GRDDL"

*** - 3. > second bullet in formal statement
Currently, this means that the "namespaceTransformation" triple has to
be expressed in the instance document, rather than in the namespace
document. From my understanding of GRDDL, and from the text that follows
in the spec, this is not right. It probably should read

"and ?NSDOC has a GRDDL result that includes"
instead of
"and ?D has a GRDDL result that includes"

- 3. > last diagram
This diagram doesn't show the timing of things, which could be
confusing. I suggest adding numbers that show the sequence: 1) NSDOC is
transformed, yielding a namespaceTransformation, 2) D is transformed,
using namespaceTransformation.

*** - 5. > second formal bullet
Same as 3. > second bullet, it seems to me that it should read

"?PROFILE has a GRDDL result that includes"
instead of
""?D has a GRDDL result that includes"

==================


Dan Connolly wrote:
> Ben, Rachel,
> 
> I have a chunk of time cleared for GRDDL spec editing
> today, so if you can get your comments in pretty
> soon, I'd appreciate it.
> 
> "PROPOSED: to publish 24 Oct http://www.w3.org/2004/01/rdxh/spec 1.120 +
> edits by DanC, contingent on editor (DanC) satisfying comments to come
> from BenA, Chime, Rachael
> ...
> so RESOLVED."
> http://lists.w3.org/Archives/Public/public-grddl-wg/2006Oct/att-0052/18-grddl-wg-minutes.html#item04
> 
> I'm working on housekeeping details like markup validity, auditing
> the references, and the like just now.
> 
>

Received on Friday, 20 October 2006 14:57:30 UTC