xml-dev - Re: Will XML change the character of W3C? Was: Re: sunshineandstandards

[Marcus Carr <mrc@allette.com.au>]

This is getting well off the beaten path, but some of these points deserve to be
addressed.

Amy Lewis wrote:

> XML's empty-element syntax is an innovation.

Nope, SGML supports the exact same syntax with a very simple modification to the
SGML declaration.

> Its avoidance of
> minimization is (to my knowledge, which is not as broad as it could be)
> unusual among SGML profiles.

Indeed, and I still resent typing in end tags that the processor is perfectly
capable of inferring. That's why I tend to use an SGML processor first if I'm
dealing with legacy data.

> It is, in sum, a well-tuned system, a
> third system effort (SGML, HTML, XML) which seems (so far) to have
> successfully addressed some of the most egregious shortcomings of its
> predecessors.

The most egregious shortcoming of all being... bad marketing? There's not much
point in presenting SGML, HTML and XML as any sort of linear progression, because
their purposes are too different. You're better off thinking of it as SGML (and
now also XML) sitting at a data or information level and HTML sitting at a
presentation level underneath them.

> "XML *is* SGML".  I've heard that song (the XML Handbook (brought to
> you by these fine sponsors!) sings it over and over and over).  XML
> isn't exactly quite a profile of SGML, as I understand profiles; it's
> more of a dialect and restriction of SGML.  To say otherwise is, I
> think, to refuse credit to a working group that really did an amazing
> job.

For the most part, I think that calling it a profile works pretty well. SGML
doesn't dictate what your element delimiters are - XML requires that they be
angle brackets. To me, that looks a lot like a profile.

Nobody's denigrating the work that was done to bring out XML - it's just that if
you had spent a few years in the small, dark, canvas tent that we fondly call
SGML, you would realise that there was a lot of it already in place. Often
modification of only moderate complexity can make something very much more
valuable than it was, but it still requires people with the vision to recognise
the scope. The contribution is in innovation and vision though, not invention. I
doubt if many of the WG would even argue that point.

> There aren't any dwarves around to stand on the shoulders of, sorry.

Perhaps we're not looking for dwarves.

> HTML was an important demonstration of some of the weaknesses in SGML,
> notably laxness in parsing as related to markup minimization.  (in
> other words, not a botch, but a grand, magnificent, *splattery* botch).

SGML *requires* parsing - if you're looking to point the finger for crummy data,
a well-formed document is a much closer relative to HTML than an SGML document
is. An SGML document has to follow all of the rules - that's why many people say
that it's too hard.

> Hmmm.  Is the thrust of this, then, a rebuttal of my derision of ISO?
> Perhaps that's deserved, although the pace of development of SGML seems
> to me to have been rather placid...

... as opposed to the pace that we're thundering toward XML 2.0? What would you
have changed and how do you feel you would have achieved consensus? What would
your reaction be to a change that made XML 1.0 incompatible? Even in XML's
relative infancy, there is no way you'd get it through. We have seen exactly the
same process occur with SML - if it's too hard to make changes, your best bet is
to rename it and start again.

> and I see no evidence that it would have evolved in the direction of
> XML without a grand stinking botch to show up as many faults in the
> existing definition as possible.

How is it that SGML gets the blame for HTML, but not the credit for XML? XML and
HTML both just borrowed concepts from SGML, and then modified them to suit what
they were trying to achieve.

> I doubt that we'd have an XML, or
> anything similarly structured and parse-able, without a W3C under which
> to issue it

We even now still have SGML. One of the overriding reasons that XML came about
was that the SGML community was worried that the HTML people were going to do
something off-the-wall. The SGML community had 20 years of dealing with
structured data and felt that the only way that HTML could go was downhill from
where it was, so they started working with the HTML people under the auspices of
the W3C. I hated the idea at the time - it seemed plain dumb and some of the
whispers indicated that there were some very bad things being discussed. For
example, I was once told that initially there was a third type of XML document
behind valid and well-formed. I believe it was to be called invalid, and it would
account for all of the crap HTML that still exists. Although there were many
concessions left in to ensure compatibility with SGML, this was given the flick
even though it would have provided compatibility with HTML. (This is not intended
as a judgement of the participants based on their background - collectively, they
came up with XML 1.0 and they all deserve the credit.) Like it or not though,
SGML contributed to preventing what might otherwise have been a disaster. Sorry.


--
Regards,

Marcus Carr                      email:  mrc@allette.com.au
___________________________________________________________________
Allette Systems (Australia)      www:    http://www.allette.com.au
___________________________________________________________________
"Everything should be made as simple as possible, but not simpler."
       - Einstein