- From: Harry Halpin <hhalpin@ibiblio.org>
- Date: Thu, 26 Jul 2007 09:39:28 -0400
- To: Bruce D'Arcus <bdarcus@gmail.com>
- Cc: bnowack@appmosphere.com, Garret Wilson <garret@globalmentor.com>, Semantic Web <semantic-web@w3.org>
Bruce D'Arcus wrote: > > Harry Halpin wrote: >> While the OWL hack is workable, again, the goal of this is to keep vCard >> as simple as possible. The only real bug in our goal is that some of the >> data in VCard, such as names, is by nature ordered. While collections >> are closed, I concur with Benjamin that we could use a container ) such >> as rdf:Seq as used in the older VCard/RDF Spec [1]. It's currently my >> preference. although sequences are not, per se, closed like a >> collection, but it's close enough for me >> >> Here's an example from [1]: >> >> <vCard:TITLE> <rdf:seq> >> <rdf:li> Principal Research Scientist </rdf:li> >> <rdf:li> Visiting Professor </rdf:li> >> </rdf:seq> >> </vCard:TITLE> >> >> So I think we could s/vcard:additionalNames/vCard:TITLE in this >> example and we'd have a pretty good solution. Garret - do you agree? >> I'd like to push out your changes to the spec right now, but want >> some agreement on this beforehand. > > I'm now totally lost. > > Everyone I have talked to has discouraged use of rdf:Seq. AFAIK, you > cannot really query it reliably with SPARQL either, though am not sure > of that. Is that true? Oh dear. Anyone have more info? > But also the original update simplified things by removing order from > the modeling so in part to make it easy to convert between it and hCard. That is a good use-case. > So what are we now proposing, for what use case? > It's not clear to me how allowing order in additional names is > actually helping us, for example. Are we also proposing it be allowed > elsewhere?? The two options are to: 1) Have all names, including "additional-name" be unordered, and then have a sort-string that orders them. That was Norm's design decision. 2) The compromise with Garrett is to allow "additional-names" to be an ordered list/sequence of some kind, as well as "honorable-prefixes" and "honorable-suffixes". Therefore, we have 3 places with names can be ordered. This is important for Garrett as he is interested in converting vCard to vCard->RDF, and preserving order in a machine-processable manner, which is difficult with reliance on just "sort-string" I would suggest that the compromise is a proper-superset of Norm's proposal, since we should not force the use of sequences/lists where not needed, as in: <vcard:honorable-prefixes> <rdf:seq> <rdf:li>Mr.</rdf:li> </rdf:seq> </vCard:honorable-prefixes> should just be: <vcard:honorable-prefixes> Mr. </vCard:honorable-prefixes> but would allow one to mark ordering where important <vcard:honorable-prefixes> <rdf:seq> <rdf:li>Sir</rdf:li> <rdf:li>Mr.</rdf:li> </rdf:seq> </vCard:honorable-prefixes> Does that make sense? I imagine hCard transforms will *not use* any Sequence or Lists, but they will not be banned by the spec, instead the spec will say precisely where they are allowed. This makes writing generic vCard/RDF processors (as Garrett I think is doing) easier, unlike the original vCard specification which allows Bags/Alt/Sequences to be basically anywhere. > Bruce > > -- -harry Harry Halpin, University of Edinburgh http://www.ibiblio.org/hhalpin 6B522426
Received on Thursday, 26 July 2007 13:39:46 UTC