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

RE: linking, 80/20

• To: 'Didier PH Martin' <martind@n...>
• Subject: RE: linking, 80/20
• From: "Bullard, Claude L (Len)" <clbullar@i...>
• Date: Mon, 19 Aug 2002 16:04:36 -0500
• Cc: xml-dev@l...

Use SGML? Well... ah heck, why not?  HTML is an 
example of that.  It is possible to overbuild any 
gencoded set.  So tear it down into modules.  Oops? 
You mean the customers won't give up the earlier 
versions because they work ok for what they want 
to do?  Ok. Take support out of the browser and 
sideline them.  Oops?  You mean they won't accept 
that and will go to a competitor if we do that? 
Well, then, we need new features they can't get 
anywhere else? Oops?  You mean they call that 
"proprietary"?  Damm.  Who sold us this markup crap?

Opennness comes at the cost of somebody supporting 
the legacies of earlier designs/mistakes/features.

Why XML?

Because we wanted a simpler markup metaspec.
XML 1.0 is simpler.   That is not the problem.  The 
problem is in the framework of applications specs that 
surround it.  SGML had very few of these and the ones 
it did have (hytime, dsssl, fosi, esis, etc.) were just 
about dry when the gencoded web exploded and forced 
a do-over.   Lots of folks processed SGML without 
DTDs before XML.  That wasn't the big leap.  The 
big leap was to accept the idea that state could 
be conveyed in XML to loosely coupled applications. 
So far so good:  but if the exchange is blind, then 
the a priori semantic description becomes a problem, 
and is usually solved as all blind men solve that 
problem; they feel around it and discover the shape. 
As in the story, what they declare it to be varies, 
but at least if they all describe it with the same 
formalism, they can discover that too.

Me:  I belive in Contract Deliverable Requirements 
Lists, or in short; tell me the name of the super 
schema that we will all be up and downtranslating 
to.  It takes time and patience to build such, but 
ultimately, they do work.   We built XML to enable
not blind interoperation, but to enable communities 
of understanding to document their vocabularies and 
use them in day to day communications.  So far, so good.

But blind interoperation, anywhere, anytime with 
anyone?  Not hardly.

len

-----Original Message-----
From: Didier PH Martin [mailto:martind@n...]

Len said:
SGML just conflated well-formedness and validity. 
Otherwise, XML works the same way as SGML.  Both 
depend on external documents ultimately.

Didier replies:
I agree Len, and the next question would be then, why use XML if it is
as complex and less versatile than SGML?

I though that the xml framework would be consistent, coherent, that
generic modules could be re-used because, in my own mental structure, it
seems to reduce the cognitive load imposed to it and finally, the
benefit I got with XML is that no external documents are needed to
process an instance (i.e. a document). Boy, we're far from that!

• Prev by Date: Re: linking, 80/20
• Next by Date: Re: Infoset chewing gum (was Re: linking, 80/20)
• Previous by thread: Re: linking, 80/20
• Next by thread: RE: linking, 80/20
• Index(es):
  • Date
  • Thread