[
Lists Home |
Date Index |
Thread Index
]
On Fri, 06 Dec 2002 10:22:01 -0800, Paul Prescod <paul@prescod.net> wrote:
> Mike Champion wrote:
> On the other hand, if the W3C invented a _particular_ binary format and said
> that the SOAP specification can _also_ be used with that format then
> interoperability would be possible and lock-in prevented. But the spec as
> it is is wide open.
> You didn't respond to my Quake UDP packet example. If I write up a mapping
> from these packets to SOAP is Quake SOAP compliant? If so, then in what
> sense does a declaration of SOAP compliance have any interoperability benefits?
I've been pondering it :-) I honestly don't know. I'm not enthusiastic
about the W3C
inventing a binary format for the XML infoset or SOAP ... I just know that
lots of
people are noodling on it, and disagree with Tim B. that it is a bad idea
in principle.
It may not be a good idea in practice; maybe LOTS more playing in sandboxes
is
needed before standards are feasible. I just fear that if the W3C treats
the idea
as intrinsically evil, then the "standards" will come from a much less
benign source.
As for Quake/UDP and the Infoset ... I personally (speaking for NO ONE, not
even
some of my multiple personalities) think that a first step might be to
draft a
version of (or complement to, replacement for, I dunno) the Infoset spec
that
tightens up things so that specs such as SOAP could be *rigorously* based
on an
abstract "XML" data model [OK, "representation of what you get when you
parse UnicodeWithAngleBrackets into something that DOM, XPath, and XQuery
can work
with"]. Such a "Son of Infoset" spec could then be a rigorous basis for a
common
data model underlying APIs, Schema languages, query languages,
transformation
languages, etc.... and perhaps one or more alternative serializations for
specific application needs (e.g. wireless).
The W3C Infoset as written does help the SOAP spec's rigor (at the expense
of readability) by replacing somewhat handwavy definitions of its XML
syntax
with specific references to XML "information items." That doesn't help
interop, AFAIK,
for the usual "a refrigerator that says 'no Infoset-compliant items inside'
is an Infoset"
reasons. I suspect that the Quake example you mentioned falls in the same category,
but don't know enough about it to assert that.
[Yet another personality seizes control of the keyboard] I'm very
sympathetic
to the "W3C has no business westling the alligators, it should be draining
the swamp" perspective about all this. I would not assert that the results
to date
shold reassure skeptics such as you (Paul P.) I do think that -- in part
because of
the inevitable hype overshoot/collapse phenomenon, and in part because of discussions
by knowledgeable people in various neutral forums -- the Web services
world
is saner and more aligned with rational expectations than it was a year
ago. But I also think that eternal vigilance and constant refactoring of
assumptions and
specificaitons is necessary to maintain the trend toward sanity.
|
