[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: SkunkLink: a skunkworks XML linking proposal

• To: xml-dev@l...
• Subject: Re: SkunkLink: a skunkworks XML linking proposal
• From: "Simon St.Laurent" <simonstl@s...>
• Date: Mon, 6 Jan 2003 20:01:05 -0500
• In-reply-to: <20030107003651.GH4661@p...>

ziggy@p... (Adam Turoff) writes:
>> I'd love to have out-of-line linking someday, but don't consider
>> XLink's approach to that particularly sensible either.  (Yes, I'm
>> working on more positive proposals.  The XLink WG is dead.  Long
>> live - or get started - XML hypertext linking!)
>
>I've heard the praise and hype for HyTime and out-of-line linking,
>but I can't say as I've seen a single practical need for it.  What
>would a sensible out-of-line linking language provide that
>SkunkLink+RDF (or the moral equivalent)  *can't* provide?

My primary problem with RDF, XLink, and their current moral equivalents
is their obsession with URIs as magic identifiers.  I've stated a number
of times that I find working with RDF quite painful because it seems to
believe that URIs are the one true identifier for everything, and so I'd
much prefer to work with a dedicated linking vocabulary - but then I
also find that XLink is not a very good linking vocabulary.

Most of my complaints about XLink arise because it crams all kinds of
things into namespaced attributes, and then it repeats RDF's mistake of
using URIs as identifiers for roles and arcroles.  The net result is
either unreadable sludge or a vocabulary that hides much of its content
behind defaulted attributes, which have their own charming reliability
problems.

URIs also have some serious drawbacks for hypertext linking, especially
if for some reason an application (typically an application performing
some kind of embedding) needs control over the content type, language,
or other media feature of the representation it retrieves.  XLink and
pretty much every other W3C specification just plain punts on these
issues:

http://lists.w3.org/Archives/Public/www-tag/2002Dec/0260.html

For inline linking, I don't mind punting  - I don't likely have the room
for precise control, though maybe it can be done in an attribute if
somebody adds the functionality.  If I'm going to go to the trouble of
setting up external linkbases and trying to build vaguely usable
systems, however, it would be nice to be able to have a conversation
about content negotiation that extends a bit beyond "throw me whatever
you like."

XLink is too complicated for the truly simple stuff and too [limited? /
naive? / attribute-twisted? / URI-blinded?] for the harder stuff.  From
my perspective, it's a good clean miss of any 80/20 point whatsoever.
-- 
Simon St.Laurent
Ring around the content, a pocket full of brackets
Errors, errors, all fall down!
http://simonstl.com -- http://monasticxml.org

• Follow-Ups:
  • Re: SkunkLink: a skunkworks XML linking proposal
    • From: Adam Turoff <ziggy@p...>

• References:
  • Re: SkunkLink: a skunkworks XML linking proposal
    • From: Adam Turoff <ziggy@p...>

• Prev by Date: Re: S-expressions vs. XML
• Next by Date: RE: SkunkLink: a skunkworks XML linking proposal
• Previous by thread: Re: SkunkLink: a skunkworks XML linking proposal
• Next by thread: Re: SkunkLink: a skunkworks XML linking proposal
• Index(es):
  • Date
  • Thread