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

RE: Re: Recognizing the contribution of the developers of XML

• From: "Len Bullard" <cbullard@xxxxxxxxxx>
• To: <xml-dev@xxxxxxxxxxxxx>
• Date: Tue, 29 Aug 2006 18:21:16 -0500

So the complaint is that SGML has a lot of features and buying a
fully-conforming system is expensive?  I can't quarrel with that.  SGML
systems were expensive and that was the real barrier to its adoption, not
its complexity.  Does anyone really believe anymore that it was too complex
or merely too complex for a Desperate Perl Hacker?

SGMLS was free but it didn't do everything.  On the other hand, what it did
was useful so we incorporated into IADS for batch parsing.   XML doesn't do
what everyone wants.  Should it?  What would it cost?   How much would one
have to spend to get a system like that?  Did the DePH vanish as a species
or was he/she always a golem used to scare the children?

I used some of the features of the SGML Declaration.  Can I get that back
too?  At what point should I punt XML away and return to using SGML systems?
Maybe there is a competitive edge for SGML vendors again.  It seems to me
that that alternative is better than a half-dozen or so XML-wannabes
competing for market share based on features that only a small percentage of
the developers would use possibly adding to the overall incompatibility of
the web datasets.

Steve Newcomb once referred to XML as an SGML Onramp.  One wonders.

Curmudgeonly yours,

len

-----Original Message-----
From: peter murray-rust [mailto:pm286@c...] 
Sent: Tuesday, August 29, 2006 5:03 AM
To: xml-dev@l...; h.rzepa@i...
Subject:  Re: Recognizing the contribution of the developers of XML

At 07:07 29/08/2006, Rick Jelliffe wrote:
>peter murray-rust wrote:
>>Remember also that XML was a direct descendant of SGML. SGML was 
>>typical first version system - over-ambitious and (I believe) never 
>>fully implemented in a single piece of software.
>There are two optional features in SGML that were not implemented by 
>the dominant parsers
>(OmniMark, SP, SGMLS) which were DATATAG and CONCUR. IIRC there were
>private systems that used them however. (E.g. TEI has CONCUR YES.)

I didn't mean to imply that there weren't good compliant 
implementations of SGML :-) merely that AFAIK there was no   (widely 
available) system that did everything in one package. And in my 
position - academic or self employed the only reasonable option was 
(n)sgmls.  I *was* interested in CONCUR and I didn't find it possible 
to find a system on which I could develop my ideas. Perhaps that is a
blessing.

>To say that SGML was never fully implemented assumes that the spec 
>was written with that
>assumption.

I didn't say that :-) ... I am aware that all parts of the SGML spec 
were implemented *somewhere* but that one might have to buy two 
implementations to have a complete range of functionality.

>On the contrary, the provision of optional features with clear 
>conformance levels
>shows otherwise. That James Clark or Sam Willmott or whoever decided 
>not to implement
>a certain optional part of SGML shows it is good to have optional 
>parts not enormous monolithic standards. So I am not aware of any 
>part of ISO 8879 that was not implemented somewhere.

I am also keen on levels of compliance in design and am trying to 
implement this in CML when possible. I think XML made a reasonably 
good decision about what was optional (e.g. validating parsers). 
MathML also has different levels of compliance. Are there other XML 
specs which have major compliance levels?

>P.




Peter Murray-Rust
Unilever Centre for Molecular Sciences Informatics
University of Cambridge,
Lensfield Road,  Cambridge CB2 1EW, UK
+44-1223-763069 

• Follow-Ups:
  • SGML complexity (was: RE: Re: Recognizing...)
    • From: Tatu Saloranta
  • Re: Re: Recognizing the contribution of the developers of XML
    • From: Robin Berjon

• References:
  • Re: Recognizing the contribution of the developers of XML
    • From: peter murray-rust