Lists Home |
Date Index |
Not at all. Well-formedness is how IADS worked long
before XML was even a proposal. It even had SGML-coded
stylesheets. That's just SGML As Practiced. It didn't
matter if XML took off. Markup of some format would
still be the way to go because in a loosely coupled
distributed system, data objects work better. People
had already tried to use OOPs for messages and that
didn't work because interoperability was not achieved.
Remember, the SGML hypertext world was intensely
experimental and most solutions had been tried.
The XML party was started by people from the print world
with the exceptions of Kimber, deRose, Durand and ...???.
There were database folks in there too, but not many
hypertext experts until the WG was formed. Then it was
a really big party.
The hypertext SGMLers were already doing this stuff.
We just fought over scrolling vs framing and whether
hyperlinks should be one way or n way, that sort of thing.
DSSSL fought HyTime and everyone fought FOSIs although
they sort of worked. What the web said, and Paul is right
about this, is This Is What An Address Looks Like. Had
it stayed there, it would have been jake. Instead, they
sprinted into the philosophical bog of universal identifiers.
All one does is make the DTD/Schema optional. That is
obvious. Using instance syntax for schemas and inserting
hidden values into the output, those would have been
controversial to SGMLers. The first would be considered an
application language just as it is now and the second
would have been a fight in the ESIS committees (what
is the output of the parse, what to do with #FIXED
values, where to define the semantics of fixed values,
etc. --- iow, same as now as Gavin just pointed out.)
XML is less than SGML. That was the idea, wasn't it?
The problem, as predicted then, was that more would
be needed. We knew that from day one.
Is that "more" XML or just XML application language
infrastructure? Consider that sending an XML Schema
in-band is just sending an application layer representation
with an instance of that application. No big deal.
Just more data objects somewhat like VRML protos.
From: Mike Champion [mailto:firstname.lastname@example.org]
3/5/2002 3:37:12 PM, "Bullard, Claude L (Len)" <email@example.com> wrote:
>Ummm... what's the difference in in-bandness in
>the instance vs just requiring a schema of some kind to
>travel with the document, other than that puts one back
>in SGMLLand pretty solidly?
Hmmm ... interesting point. Having a travelling schema
would address my original concern.
>We do seem to work a lot to preserve well-formed options.
Well, well-formedness brought us to the party, IMHO. We
would be a lonely group of SGMLers whining about the incompatibilities
between Word and HTML, and CORBA and DCOM, and wishing that the
world would just take a sip of our KoolAid -- XML would have
never "taken off" without the WF "foot in the door." I suspect
you disagree :~)