RE: [xml-dev] Re: determining ID-ness in XML

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

Reaaonably, yes.  Realistically, I can't say.  Language 
designers for the WWW are free to do as they will and 
that is precisely the problems the Hytimers were tackling. 
Why?  In their world, they didn't see a system unified 
by syntax, but by agreements on abstract forms each 
notation processor had to support.  That was a tough 
love approach, but it had some promise.  Again, what 
does an XPointer do with a PDF document?  Nothing, 
probably.  Could it?  Possibly yes.  It requires 
agreements that are somewhat abstract and that is 
why guys like Dan Connolly said, "I just can't grok 
HyTime."  It wasn't simply obscure; it was saying something 
that it takes time and experience to understand; some 
forms of control really are doctrinal; they tell you 
what you have to do (the intent, the results), but 
not how.  This is a lot like trying to teach the 
infoSet to a room full of COM programmers who keep 
looking up and saying, "show me the code! show 
me the code!"  They are usually a year or so into 
their development before they realize they don't 
know how the system works.

IDness is the root fundamental of reliable 
addressing.  Tim and James are right to point 
out this is a gap, but it is a gap the XML 
WG and SIG put there.  So now we have to ask 
ourselves how much of this we want to solve. 
My feeling is that given the current proposals, 
we will be extending the system vocabulary 
every few months.   I am also concerned that 
regardless of the feeling that XMLIsTheThing, 
the other hypertext notation targets will have 
the same issues leading inevitably to the 
reinvention of Hytime under another name and 
that we may have to come back and do this 
again with even higher costs.  This is how 
80/20 bites us.   We can punt it away and 
solve only part of the problem now, and that is 
what we are likely to do given the politics 
and personal persuasions.  The Web Architecture 
group should stop and ask itself if traditionalism 
works against effective design.

len

-----Original Message-----
From: Leigh Dodds [mailto:ldodds@ingenta.com]

Something else to consider:

CSS provides a Selector based on ID [1]. 

The text at [1] acknowledges that ID linking is unreliable 
(DTD might not be processed or might not exist), and presents 
a work around: using attribute selectors.

Looks to me that any solution to the 'ID problem' will have to work 
with CSS, and presumably impact the CSS specification as well, yes?