"uid" microformats? (was Re: [uf-discuss] ISBN mark-up)

[Tantek Ç elik]

On 4/24/06 7:42 AM, "Bruce D'Arcus" <bdarcus.lists at gmail.com> wrote:

> On 4/21/06, Tantek Çelik <tantek at cs.stanford.edu> wrote:
> 
>> Just as we extracted "geo" and "adr" from hCard to address the use cases
>> brought up the Geo community for a simple format for locations, perhaps we
>> can extract "uid" from hCard for this purpose and be done with it without
>> adding any additional vocabulary.
>> 
>> Thoughts?
> 
> A question, really: what are the semantics of uid? It's not clear from
> the hCard spec,

It's quite clear. hCard normatively references vCard for semantics, which
says:

3.6.7 UID Type Definition

   Type purpose: To specify a value that represents a globally unique
   identifier corresponding to the individual or resource associated
   with the vCard.


> and poking around on uid and vCard suggests that it is
> a concept that may well predate uris?
> 
> If that's the case, I wonder if it doesn't make more sense to call it
> uri, and bring that back to hCard?

If you believe there are non-trivial advantages (especially to publishers),
go ahead and add it in a new section on hcard-brainstorming so we can at
least capture the proposal.

http://microformats.org/wiki/hcard-brainstorming


My intuition says this is an unnecessary change.

First, in general changing names of properties in microformats is a bad
idea, unless you are doing so to reuse from other microformats and thus
reducing the overall vocabulary (which we did in a revision of xFolk).
Nothing is using "uri" currently, so this would actually be an expansion to
the vocabulary.

Second, changing the name of something from the standard which you are
attempting to interoperate with (vCard) is also a bad idea (unnecessary
namesmithing), which results in lots of subtle name changes (see XML/RDF
attempts at vCard/iCalendar) which were made essentially at the whims of the
authors of the specs (I have found no documented methodology for the
changes).

In microformats, *every* change from the name in a standard we have
documented meticulously, and documented the methodology for doing so (e.g.
plural vs. singular, explicit sub-property names taken from the prose
descriptions in the standards, etc.)

Third, I actually see disadvantages in using URIs as a basic unit rather
than URLs.  URLs are far more useful in that they assert you can go get this
thing, it has a location, most likely on the Web.  Thus as a pattern we
should use URLs in microformats, not URIs.

OTOH, an opaque UID which asserts nothing but "globally unique identifier"
(see above) is both quite simple, and much more "backwards compatible" with
use of UID in vCard/iCalendar applications today.

Thanks,

Tantek