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


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: SV: [xml-dev] Tim Bray on "Which Technologies Matter?"

[ Lists Home | Date Index | Thread Index ]

Hello Daniel

Daniel said:
>   So everybody agreed on an SGML implementation ?
> Either choice is good or not, you cannot defend both. And as far as
> I understood there wasn't that many SGML processor (especially for
> a reasonable price) though people seems to have suffered long and
> hard that each processors had his own set of "features".
>   So which one were you all using ;-) ?

Didier replies:
When you talked about an SGML implementation you probably meant here a
parser. If that is the case, most of the actual SGML tools where based on
James Clark's parser (i.e. SP) and most recent ones on its successor
(OpenSP, ref: http://openjade.sourceforge.net/). If that wasn't the case,
and you meant an SGML based language (i.e. a domain language), obviously, as
you know HTML is one but also, as others previously mentioned it, the
official aircraft documentation language named ATA 100 (and its variants) is
probably SGML's biggest success. Again, the real issue is not about the meta
language per se but more about the derived language acceptance and
standardization. The fact is that actually the most popular rendering
language is based on an SGML DTD not an XML schema (I'm talking here of
HTML), however, people don't even care about the fact that HTML could be
used to structure documents or that a good practice is to separate the
content from the views. HTML is mostly used as a rendering language, and
people used it because it was easy to learn, and tools became available
early (most of them though are not using SGML parsers nor are using XML
parsers today). It seems that people care more about the domain language
than about the meta language used to create it. And if you talk to people
using the ATA 100 specification  to publish aircraft documentation you'll
probably notice the same fact.

Now a bit of objectivity:
We cannot say that in the XML world all parser implementations are
homogeneous and supporting the same feature set. However, we can say that
more domain languages have been created with XML than with SGML. Using the
Geoffrey Moore's model (ref:
http://www.chasmgroup.com/aboutus/mds/moore.htm) we can say that XML crossed
the chasm and is slowly being accepted by the main stream, on the other
hand,  SGML never crossed it but got some success in certain niche markets
(i.e the aircraft market). Or if we are good observers, we can say that the
web helped SGML to morph into XML. During that process, we saw the
appearance of SGML for the web and softquad panorama was an early prototype
of an XML browser. Then, after several draft about a simplified SGML (mainly
to fulfill the needs of the web), XML came out of this refinement process.
Again, if we are good historians, we can say that the result of the industry
learning and adapting is the simplification of SGML into XML. Ideas could be
like species, they evolve and adapt to their environment. It is the survival
of the fittest. Of course we can ask: are the ideas modified because of the
environment or are the ideas simply accepted or rejected by the environment.
Maybe a bit of both.

And the whole thread make me use some glucose to feed starving neurons and
come to some more thoughts.

Actually, if we are still good observers, we can say that XSLT is making
progress, XLINK is lost in the no man's land, XSCHEMA is controversial, and
so on and so forth. Some rendering languages are making good progress: WML,
VoiceXML, SVG. It seems also that XDR and tutti quanti RPC marshalers could
be replaced by SOAP. Especially if we consider the billions Microsoft and
big Gorillas are investing in it. Thus, SOAP may help the industry do what
CORBA, RMI or DCOM couldn't do before: a common RPC marshaler. However, it
won't resolve anything else since it won't standardize the API's. Again, I
am still amazed to see how Bill's "Embrace and expand" is among the best
conqueror's strategy  I saw since a long time. Said differently, the
marshaler is standardized but the API is proprietary. And the API is
precisely what could bring the power to the buyers (ref: porter's
competitive model - http://www.brs-inc.com/porter.html).

In my humble opinion, the domain languages are more important than the meta
language. Agreeing on the alphabet do not help us speak the same language.
Just consider that most European/American language are roughly using the
same alphabet - can you be understood everywhere in America or Europe?
What's important is no longer XML per se but what we do with it. We agreed
on a common alphabet (i.e. XML) now we should agree on the languages based
on this alphabet. Having a couple neurons to recognize sound pattern is a
good thing but it is even better to have some gray matter to process it. So
the real question is: what do we do with this alphabet? Where is the power
located? On the vendor's side or on the buyer's side? As long as we focus on
the alphabet it will be on the vendor's side. If we focus on the languages
created with the alphabet and ask the vendors to conform to it, the power is
now in the hands of the buyers.

Didier PH Martin


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

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