[uf-discuss] Almost UMP:xFolk Supported

[C. Hudley]

On 12/8/05, David Janes -- BlogMatrix <davidjanes at blogmatrix.com> wrote:
>
> "@uri" and in fact "@anything" are little pieces of information the
> parser has picked up among the way, not necessarily part of the uF. In
> this particular case, it's the uri of the page being parsed.

It would be nice to be able to pick up per-microobject URIs, should
they be specified.


> There's another I added this morning, called '@index' which allows the
> uF element to be corresponded to the parsed version. For example,
> '@id=vevent-3'means that the 3rd element with "vevent" in the class
> attribute is the one that was parsed.

That's very helpful!


> > Over on gcs-pcs-list we've been talking about a simplification of the
> > COinS spec when simply indicating what-the-identifier-is, which is
> > (obviously) rather lost inside the OpenURL ContextObject inside the
> > span.  Lately we've considered instead just using "class='uri'" and a
> > URI in a title or as the textnode content.  I'm experimenting with
> > something like this in the canary database as well.
>
> How about rel="bookmark" (i.e. rel-bookmark) Not only is it used in
> hAtom, it's also part of the HTML spec [1]

As I understand it, rel="bookmark" implies that identified items
are/have directly resolvable http URLs.  In our use cases the URIs
often include not-directly-resolvable URNs, such as those constructed
using the recently registered "info" URN namespace id (e.g. ISBNs or
DOIs... and "just point to amazon or the DOI foundation" is not a
sufficient solution :).

I'd presume - perhaps incorrectly - that most applications parsing
rel="bookmark" items would in turn assume that they are http-locatable
URLs.  This is certainly the case in hAtom, where it's specified as
such (unless I misunderstand the concept of "permalinks").  If that's
the case, then this wouldn't work.