OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: "XML is a subset of SGML" - True or False?

[ Lists Home | Date Index | Thread Index ]
  • From: Rick JELLIFFE <ricko@geotempo.com>
  • To: xml-dev@xml.org
  • Date: Wed, 05 Jul 2000 18:00:45 +0800

Ralph Ferris wrote:
> >--- In xml-dev@egroups.com, Rick JELLIFFE <ricko@g...> wrote:
> >
> >Where there were differences,  ISO8879 was corrected specifically to
> >make sure that XML was indeed a subset. In fact, Charles Goldfarb
> >even said at one stage "XML *is* the revision of SGML" (debate on the
> >revision of ISO 8879 had started years before: XML was the embodyment of
> >that).
> This message, while technically accurate, also contains a large element of
> the disingenuous. It almost makes it sound as if Charles Goldfarb was the
> initiator of the work on XML, as the logical next step in the evolution of

Since "disingenuous" is a serious charge, I hope XML-DEV people will
forgive a longer answer, but I suppose most people on XML-DEV should be
interested in neither Ralph's response nor my rejoinder here.

When I put up a grammar on comp.text.sgml in about 1993 or 1994 for
SGML-Lite, Charles contacted me and he told me that the big criterion
for revising SGML had to be that the changes could be demonstrated to
reflect user-driven/vendor-acceptable demands: anyone can come up with
nice proposals to simplify or improve SGML, but if they have no
consensus in the community, they don't have credibility. I took it as a
challenge that a grass-roots movement was needed, which was completely
beyond my abilities or ambition. So the fact that the next step in SGML
should be the result of extensive community debate and consensus outside
ISO then presented to ISO, yes that is exactly how I saw what Charles
told me 6 years ago: the means and the form and the impetus and the
completion and the will of course of course fell to Jon Bosak.

So the heart of the delays in the revision process were not that the ISO
committee was being bloody-mindedly slow or delaying, but that the SGML
user-community & vendors could not get together to produce sufficiently
worked out plans in detail. 
Ever since XML came out people have been trying to mythologize that it
represents the triumph of the little man over the faceless ISO men; yet
as I have pointed out the faceless ISO men were in the same people as
the XML IG and WG. 

I strongly object to the claim of being "disingenuous": I think Charles
Goldfarb's democratic position in this has been clear, consistant,
frustrating and correct--changes to SGML (e.g. XML) (that are not
corrections of errors) should come bottom-up from the user community:
the ISO committee should act to rubber stamp (i.e. to formalize in
consistent terms) the devlopments in the living markup community. As the
W3C effort on XML proved itself over time to be a legitimate phenomenon
in this, Charles participated with no antagonism that I could see (e.g.
he suggested the resolution of how to treat empty end tags as NETC for
SGML, he even suggested getting rid of mixed content at sone stage--that
is hardly the action of someone trying to hold things back
desparately).  When we had the ISO WG that created ISO 8879 Annex K and
L, I suggested Annex L as the appropriate example of an Additional
Requirements document and as a way to clearly show that profiles such as
XML in general and XML in particular should be considered "mainstream"
use of SGML.

And please don't mythologize that this represented some major post facto
change in thinking from Charles: in my book The XML & SGML Cookbook,
which Charles edited, I had a section on simplified versions of SGML
which was written a year before XML's release and on which Charles did
not demur (of course, Charles' series does nto represent his opinions:
he restricts his series-editorship to conceptual and terminological
consistency and bringing out important ideas, not on making his authors
tow the party line.)

Charles' position reminds me of the Pope: when asked "Why are you a
conservative?" he said "But my job is to conserve!"  I think everyone
who has ever developed a public spec. will have sympathy for this: at
the publication date one suddenly changes from being (regarded as) a
progressive path-setter cutting through the crap to deliver something
new and real, to being an old fuddy-duddy reactionary standing in the
way of progress--the establishment, a member of a conspiracy against the

If we took the XML spec and looked at each part and said "who invented
this tag or concept?"  we would not get Jon Bosak or almost any member
of the XML WG for almost any of them. Perhaps "/>" and the XML header
and the required target for PIs are new (using no DTD has been common
practise in non-standard SGML systems for ever): but apart from them,
calling Jon Bosak the "father" of XML is to to some extent to ignore its
genuine intellectual provenance: Jon is certainly the foster father or
midwife or leader or marshall or organiser, XML would and could not have
happened without him and XML is the best thing to happen to SGML. Jon is
the great hero for this technology, and a remarkebly nice and capable
guy.  But when I read people on this list vituperating because a
newpaper article claims MicroSoft invented XML  but then replacing this
fallacy with another, why shouldn't I respond?

As far as my claim that XML is the embodiment of the debate on the SGML
revision, note that XML includes several features already slated for
support in the revision 
 - HEX NCRs 
 - name characters for almost all Unicode characters 
 - more emphasis on general linking rather than entities
and that the debates on XML were completely informed by the years of
going through various SGML issues by the ISO working group.

Charles' statement to the effect that XML is the SGML revision came
sometimes after XML 1.0 or Annex K & L were put out.  So certainly I
apologize if anyone thought I meant that Charles was directing or
instigating XML, to the expense of giving Jon Bosak his due credit.

But that is not being disingenuous.

> The reality is, the 10-year revision of SGML was proceeding at a glacial
> pace, when Jon Bosak perceived the need for a "global warming" to save SGML
> from being permanently marginalized. And at a time when Netscape was riding
> high with no interest in HTML as an "SGML application", that's exactly where
> it was headed.

The trouble with this statement is the idea that the speed of the
10-year revision was the problem: in fact, the things mooted for the
revision would by and large moved SGML into the direction of more
complexity not less. The Japanese mopdularization proposal is a good
example: it is excellent technological, but it is not a simplification
that would have won the heart of Netscape.    I am not saying that the
slow speed or the difficulties in justifying changes was not what caused
Jon to start XML, I have no knowledge in that; however industry groups
provide better forums for creating profiles than standards
groups--standards groups are interested in "does this actually solve the
general problem" while industry groups are interested in "what is the
80/20 point" or "how much do we need for our specific application". 
What Jon's perceptions were is a different question to what actually
caused things to pan out so well: with the benefit of hindsight I would
go as far as suggesting that the slowness in getting the 10-year
revision has saved SGML from bloating further.

> >In fact, XML is explicitly part of ISO 8879, through non-normative
> >Annex L: XML is used as an example of an "additional requirements"
> >document--it cannot be normative because the ISO WG thought that W3C was
> >an appropriate body to create and maintain an industry-lead profile of
> >an ISO standard.
> Again, this makes it sound as if ISO, in it's infinite wisdom, graciously
> decided that the XML work should go forward within the W3C. 

No, I don't think it does.  Perhaps  s/thought/recognized/ would be
clearer and 
s/an/the/g would be congenial.

One reason why some people want to deny that XML is a subset of SGML is
because they want to allow W3C, an institution paid and dominated by the
largest companies in the world, to be able to alter XML willy nilly as
it suits them.  Of course, ISO committees are also dominated by
commercial concerns, but to a far lesser extent.

To avoid any charge of disingenuity here, I will be very frank in my
views: in the world of standards there is basically only the choice
between commercially-dominated groups, Western/English-dominated groups
or ISO and national standards bodies. The idea that somehow W3C is a
good ground for simplicity or "technology for the rest of us" is belied
by the facts: look at the current generation of WDs, CRs and LCs.  (This
is not a criticism of W3C.)

Rick Jelliffe
(writing privately)

This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS