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

RE: Validation - a history.

• From: "Len Bullard" <cbullard@h...>
• To: "'Dave Pawson'" <davep@d...>
• Date: Fri, 2 Mar 2007 07:32:00 -0600

1.  The "NO DTD" crew lost.  There was too much experience with DTDless
systems at the time to let that happen.   The optional DTD crew won because
we had plenty of experience with that.

2.  The cost of an SGML parser had already gone to zero by the time the SGML
ERB was formed.  We were using SGMLS in IADS by that time, and had used
Goldfarb's free parser before that.   The high cost of parsers had caused
that source to become available.  Charles was extremely generous on that
point.  Clark worked with that code to create SGMLS.  I don't know if
Charles intended the source to be used that way, but I don't recall him
objecting to it other than to say that he had to teach himself C in a
weekend to write it.  It might not have been Clark-quality code but a
technical idiot can't do that and a closed mind won't.

3.  It is rarely happening because of ten years of large scale practice and
plenty of samples.  Prior to XML, SGML was used by major system vendors and
they did not give up their DTDs to the public easily.  The most common
example was the military adaptations of 38784 that became MIL-28001 and it
was over a thousand pages of DTD because the thinking was One Size Fits All
so they made it *comprehensive*.  That kind of thinking was the real problem
with DTDs, not that they couldn't be made to work in hypertext systems.  The
NIST guys would make claims like that and we would show them IADS working
with optional DTDs.

4.  Yes, WGs are often filled with conflicting points of view sometimes
because the extremes are met with extremes until a middle point of view that
meets the requirements emerges.  Years later, the winners of those conflicts
are often shown to be wrong.  The No DTDs crew was dead wrong.  The optional
DTDs crew left room for the emergence of various schema languages.  SGML
Declarations allow for developments like JSON and an Infoset concept (The
concept was around but it wasn't called that) are an improvement on that,
but tossing out the Declaration was a battle the No DTD folks won.  Now
something like it has to be reinvented.  We were better off with a frozen
declaration as Clark provided, but the idea that something like it would
never be needed was hubris and part of the "We Are The Web The Web Is All"
thinking that forces innovation to go off the web and away from the W3C to
happen.

Things improved as they will in a technology that is heavily used and more
people get a voice.  Speeches that justify the mistakes or the triumphs at
the expense of the pioneers don't help.  They just make people feel better
about themselves.  In hallway conversations, ok.  In keynote speeches at
major conferences?  That requires rebuttal.

Pong...

len


From: Dave Pawson [mailto:davep@d...] 

Len Bullard wrote:
> That's an interesting point of view, Dave.
> 
> SGML with a DTD worked too.  Those value setting properties made parts of
> the trinity work.  That was important to the point of view that Charles
was
> arguing for at the time.  Since the poster decided to emphasize the
debates
> between Dr. Goldfarb and Jon et al, it seems fair to remember what it is
to
> be an ISO standards editor:  scrupulous correctness.  Also, since
arguments
> like "JSON is Just XML with curly brackets", it is fair to point out that
> the SGML Declaration was used to declare delimiters, not the DTD per se.

Two points. The high cost of a compliant SGML parsre and the 'discovery'
of well-formedness seemed to impact the view of the group.

> 
> XML without DTDs do work just as SGML without DTDs worked in the sense
that
> as long as there were other means to explicitly declare the system
specific
> semantics of tags (think, href is a hardwired attribute used by the Web
> System), then it doesn't matter.  
That's a given. Unknown elements will still screw a system today without
a must-ignore policy or similar. The reality seems to be that we are 
often dealing with repeats though where this rarely happens?

> But frankly, I hate to see Jon and others grind on Charles like that.  He
> pulled the train and kept it going when the rest of us were still trying
to
> figure out what a f**kin' hyperlink was.  That's classless. He's a good
man
> and he did heroic work that the rest of us are still benefiting from in
ways
> large, small, and very profitably. Since they choose to do it in public
> speeches quoted here and elsewhere, I'll grind right back and that's fair
> too.  

I didn't read it as having a go at Charles.  Just telling what happened.
He was in a minority on the no dtd debate.  That happens in any WG.

• Follow-Ups:
  • Re: Validation - a history.
    • From: Gavin Thomas Nicol <gtn@r...>

• References:
  • Re: Validation - a history.
    • From: Dave Pawson <davep@d...>