Re: Topology: XLink CR03062000

Hello again,

At 09:06 AM 11/7/00 +0000, Hartmut Obendorf wrote:
>Good Morning,
>
>I would like to add a third posting to explain one of the reasons I
>feel "complex links" (with more than one arc) to be important and
>why I think this should be mentioned somewhere (if not in the spec
>itself).
>
>One of the main problems of browsing applications for the W3 is the
>missing predictability of outgoing links. Supplying additional
>information such as file size/types, in-site vs. out-site location
>etc. could be collected with HTML links. But one thing that is very
>difficult to extract is the topology of links.
>
>Let me state an example first: Rsrc D, E and F share a common idea,
>they are linked serially. Now Rsrc A has something in common with
>Rsrc E (both are vocals) so it is linked to E. There is no way of
>extracting the information that D, E and F are related, while A, E
>and F are not (that is, A and F are not).
>
>picture:
>
>   +---+
>   | A |
>   +---+
>        \
>         \
>   +---+  +---+  +---+
>   | D |--| E |--| F |
>   +---+  +---+  +---+
>
>If, instead of HTML anchors, XLinks were used to link the resources,
>one XLink could contain the whole chain of arcs D-E-F (the other link
>would just contain an arc from A to E). So, it would be very easy to
>find out that D-E-F and A-E are closely related while A-F are not.
>
>This idea can be easily extended to keep structures like chains,
>rings or whole hierarchies in one link, thus providing structural
>"outlink information" that could be used for users to improve the
>predictability of link targets.
>
>Of course, this will not be new to you, but isn't it important enough
>to mention it somewhere?

You know, this is something that's been bugging me for a while. :-)  The 
extended-type element obviously allows for the "branding" of a set of 
relationships with role and title information on the link as a whole, but 
nowhere do we explicitly motivate the use of one vs. several linking 
elements to hold a given set of arcs.

As you mention above, I'm not sure this sort of discussion really belongs 
in the spec, but it could be the subject of a note or paper or something...

         Eve
--
Eve Maler                                          +1 781 442 3190
Sun Microsystems XML Technology Center    eve.maler @ east.sun.com

Received on Wednesday, 8 November 2000 19:23:32 UTC