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

RE: limits of the generic

• To: xml-dev@l...
• Subject: RE: limits of the generic
• From: "Bullard, Claude L (Len)" <clbullar@i...>
• Date: Mon, 30 Sep 2002 15:35:00 -0500

Ok, so you are displaying it with a stylesheet 
and what you want is for links to work:

o regardless of stylesheet

or

o by recognizing the form you have from the stylesheet.

and of course, you don't want a hyperlink to be anything 
but a control for launching a display.   It should never 
be an abstraction of "relates to" or "is a".

Think of this exercise as if HTML had never existed and you 
were being asked to develop a hypermedia system that 
will run on any platform across a network.  If you 
had to engineer that from scratch, you might start 
by trying to figure out what all of the possible 
documents have in common, and after clearing away 
all the red herrings (eg, structural items like 
paragraphs and headers and tables), you would 
come down to this one control called a hyperlink. 

1.  One side will start arguing that this is 
just a presentation issue.

2.  One side will argue that a link is a link is 
a link and that all presentation semantics are 
late bound so hyperlinks are really abstractions 
or arc/node thingies.

Then you will start building up a framework of 
objects that can recognize any link anywhere 
by some magical induction process.  When that 
gets to cumbersome, you'll simplify it to have 
a flag on it to tell you it is a hyperlink.  Someone 
will say that doesn't allow for both 1 and 2, so 
you'll add some more discriminating flags for 
roles, etc.

Eventually someone will get tired of all of this complex 
stuff and say, "hey, let's just gencode it, name it 
anchor, and declare by standard that any document anywhere 
that works in our system shall use <anchor />.  By the 
way, our system shall be the only system available, and 
that will be because it is the standard.

Y'all really are well on your way through ISO 10744 
and out the back door to HTML.   Time to make your 
requirements more explicit than a use case.

len

-----Original Message-----
From: Micah Dubinko [mailto:MDubinko@c...]

Joe English:

>How is blind recognition of links any more useful than
>blind recognition of any other data type (which is to say,
>not very)?

How about a use case:

I want to put, say, DocBook on the web. I can do this today and use a
stylesheet to make it display reasonably, even though none of my browsers
have any built in knowledge of DocBook.

I also would like the links to work.

• Prev by Date: Re: Datatypes - it's in the contract
• Next by Date: Are hyperlinks presentation or content?
• Previous by thread: Re: limits of the generic
• Next by thread: Emacs; how to override built-in xml-mode?
• Index(es):
  • Date
  • Thread