[
Lists Home |
Date Index |
Thread Index
]
On Fri, 30 Apr 2004 jcowan@reutershealth.com wrote:
> Robin Cover scripsit:
>
> > Are you forgetting about some of the
> > untested SGML 'FEATURES' that got into ISO 8879, and by common
> > consent (from my POV) represent engineering monstrosities?
>
> Something I'd like to see: a historical note explaining the features of
> SGML that didn't make it into XML, particularly focused on what they were
> intended to be used for.
I'm not sure this note will ever be written, as it would likely not be
flattering to any of the people who deserve credit for making the
core ideas of SGML a success. Read between the lines in Steve DeRose's
SGML FAQ Book, and read (twice) through the postings of Erik Naggum
to comp.text.sgml, and ask some of the people who witnessed the ISO
process at work in the final months before 8879 became cast in steel.
I could write what I think I know about this, but I don't think it
would serve any interest other than historical, and it would represent
disproportionate focus on a part of the story that's not so pretty.
One example: we know a lot about the 'CONCUR' problem (some
refs at http://xml.coverpages.org/hierarchies.html ) but the experts
I know will tell you that the SGML CONCUR feature did not solve the
"real" concur problem, and arguably, not even the problem it tried
to solve, because of ambiguities [things not/under-specified] in the
standard.
My intent was not to discredit the 8879 Standard, nor to discredit
the principal designers (most of whom were not formally trained
computer scientists, as has often been observed), but to offer one
small example showing how the ISO process itself does not guarantee
QA. That's important if the notion of "guaranteeing a better chance"
is fundamentally and profoundly non-determinative.
Robin
> --
> "May the hair on your toes never fall out!" John Cowan
> --Thorin Oakenshield (to Bilbo) jcowan@reutershealth.com
>
|
