RE: [xml-dev] Re: AF and namespaces, once again (was Re: [xml-dev] There is a m eaning, but it's not in the data alone)

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

Regards my response to Rick, it appears that you have 
concerns which, when I sort away the technical details, 
are identical to the ones expressed by various XML-Dev 
list members:  how to express semantics for XML documents. 
You go a bit further and require multiple semantics and that 
the instance should be able to tell the system which semantics 
it can be used with.  

I'm not convinced this is a struggle with large corporations, 
or that we have to keep that issue up front.  Life is doing 
that anyway and I can't think of a person here that doesn't 
express that worry here from time to time except possibly 
Clark.  I think it is the Pogo thing and the people who 
have to be convinced are well-represented here.  If the 
developers are sold on the idea, it usually sells elsewhere. 
XML was sellable because it was easy to explain and came 
with a built in "up the noses of our ugly parents" rhetoric. 
If we can achieve the first without the latter, the debates 
will at least be quieter.  What is most important is to keep 
the focus on the layering aspects, single contact as you say, 
and to make sure the requirements are adequate to produce a 
system that has at least a five year life cycle.  

It seems to me that you should be attending the ISO committee 
that James is chairing with Rick, Ken etc.   Even if AFs 
aren't the final solution (if there ever is such a thing), 
the lessons learned, the facts about the implementations 
you have worked with, all seem relevant to that effort. 
And having served with you in at least one or two 
short sword battles, I know that if the knives come 
out, you are more than capable. ;-)

But if the last month of cross-topical threads here 
are indicative, the issue is very much on all of 
our minds and agendas.

len

-----Original Message-----
From: Steven R. Newcomb [mailto:srn@coolheads.com]

"Bullard, Claude L (Len)" <clbullar@ingr.com> writes:

> Steve, are they?  Would archforms work with RNG?

Exactly what I'm wondering, Len.  I think so, but I'm
not comfortable enough with RNG yet to commit myself
on this question.

I like RNG a whole lot, though.  And, the idea of
making a clean separation (or, rather, a single point
of contact) between semantic and syntactic validation
is very consistent with the goal of modularity -- the
same goal that is at the heart of the architectural
forms paradigm.

Over the last few years, I've talked about AFs with
Makoto Murata from time to time, so I know he's well
aware of the issues.  As for James Clark, he
practically wrote the book on architectural forms,
although he professes no great love for them now.  So I
have high hopes that the requirements that I keep
bringing up, year after year, will someday be addressed
by people who are competent to address them.

I think it would be higher on their agenda if more
people indicated to them that these are serious
requirements with serious economic consequences.  Right
now, it's still unusual for someone to bring up the
possibility that a single document could be useful in
multiple processing contexts.  (And when someone does
this, typically someone else says, "You mean CONCUR?",
whereupon everyone present makes a rude noise.  The
CONCUR feature of SGML was an earlier attempt to
address some of the same requirements.)