Re: Profile URIs (and how microformat mixing can go wrong)

This message has been in my inbox, flagged for response, for months...
I'm not sure I've got bandwidth to answer it in detail now,
but microformat mixing came up again in our telcon last Weds...

jjc: one could make one profile for all microformats, or for all the
ones that we have XSLT for
DanC: yes, that just needs a maintainer to set/meet expectations.
http://lists.w3.org/Archives/Public/public-grddl-wg/2007Apr/att-0014/04-grddl-wg-minutes.html#item12

This is the sort of question a maintainer would be expected
to answer, if only to say "bummer."


On Wed, 2007-01-24 at 11:01 +0100, Danny Ayers wrote:
> On 24/01/07, Dan Connolly <connolly@w3.org> wrote:
>         
>         On Tue, 2007-01-23 at 19:35 -0500, Harry Halpin wrote:
>         [...]
>         > > ... though some social/trust issues remain.
>         > >  http://microformats.org/wiki/profile-uris
>         > >
>         > > Oh... I guess some microformats-mixing issue remain, but
>         those
>         > > aren't really GRDDL issues.
>         >
>         > I'd actually like to see a good example of how microformat
>         mixing goes 
>         > wrong.
>         
>         Here's a sketch...
>         
>         <div class="vevent">Let's have a
>           <a class="summary url"
>         href="http://example/hodown">ho-down</a> at 
>           <div class="location vcard">
>             <a href="http://example/flippies-restaurant">flippy's</a>
>           <div>
>           tomorrow 
>         </div>
>         
>         The event should have an url of http://example/hodown but
>         not http://example/flippies-restaurant . I asked tantek 
>         about it, and he said that an hcalendar parser should somehow
>         know not to pick up the url inside the vcard. I don't know
>         of any code that is that smart; in fact, I think the
>         hcard-parsing
>         page doesn't even specify how this works.
> 
> Sorry Dan, could you please clarify how the event might end with the
> wrong url - which transformation(s) are you applying here?
> 
> 
>         Then, from the GRDDL perspective, there's the issue of using
>         separate glean-hcard.xsl and glean-vcard.xsl transformations:
>         norm's glean-vcard produces a bnode for the card, so the
>         location field of the event can't refer to it. That's
>         probably 
>         fixable by using an id for the location, but we haven't worked
>         out the details yet.
> 
> That sounds like it might be rather fragile. Are any of these
> relationships inverse-functional?
> 
> 
>         >  I'll e-mail Liam Quinn over it, as I remember he brought up
>         an example at a
>         > talk at Extreme Markup Languages last year [1], but I cannot
>         find his
>         > precise example online. I do think microformats-mixing is
>         out of scope of GRDDL
>         > WG.
> 
> But if @profiles are in place then won't this mean that GRDDL doesn't
> quite work when it could/should? That would seem in scope. I don't
> think we should go into specifics, but perhaps we can identify the
> kind of circumstances that lead to (unexpected) incompleteness or
> misinterpretation. 
> 
> 
> 
> Cheers,
> Danny.
> 
> 
> -- 
> 
> http://dannyayers.com
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Friday, 6 April 2007 16:17:01 UTC