[
Lists Home |
Date Index |
Thread Index
]
On Thu, 5 Dec 2002 12:48:47 -0600, Bullard, Claude L (Len) <clbullar@ingr.com>
wrote:
> ++1.
>
> Here is the exemplar for that kind of thinking: [ a quote from Martin
> Gudgin]
>
> "Philosophers have often asked the question, 'If a DOM tree sprouts, grows,
> withers and dies and is never serialized using angle brackets, is it
> XML?'
[snip]
<non-concrete-fuzzy-analogy CanIgnore="true">
Well, philosophers used to ask the question, 'is light was a wave or a
particle.'
The answer AFAIK is "both" or "neither" or "mu" or "I could tell you, but
you'd
have to understand quantum electrodynamics, and THAT would kill you".
I think we'll come to the same point with respect to the XML syntax and
infoset (very broadly defined) someday. Gudge is right, Len is right, both
are right, and all are wrong ... there is an "XML QED" out there somewhere
waiting to be discovered :-)
</non-concrete-fuzzy-analogy>
>
> The urge to muddy what is platform and what is markup is a terrifying
> problem.
And Tim Bray said "it would be disastrous if you were able to advertise to some
external party that you provide XML, and then offered anything but
unicode-with-angle-brackets." Well, I don't think it's either
terrifying or disastrous to have a menu of standardized infoset
serializations
available -- e.g., things like WikiML for human authoring, something like
WBXML (what little I know of it) for constrained devices, something like
serialized
SAX events for high performance environments. I would totally agree that
either
calling an alternative "XML" without qualification would be stupid, but
presumably
this is a problem that HTTP content negotiation could handle easily if
someone offered
an alternative *standardized* "XML" serialization on their website.
I see several scenarios for the future: One is that the alternatives become
moot -- Office 11 makes
XML GUI authoring ubiquitous, and Moore's Law makes the bandwidth and
performance
issues associated with the XML 1.x syntax trivial. Two is to legitimize the infoset-
centric view of "XML" and have a manageable number of alternative syntaxes
(with MIME registrations, etc. etc. to keep the Web sane). Three is that
people take option 2 without the blessing of the W3C, simply because they
can't afford the
overhead of the unicode tags, etc. in their constrained or high-performance
environments,
and we have real chaos. I guess Four is that the Next Big Thing that
replaces XML for the "data" world finds a better balance between human
readability and machine processability.
People here seem to believe that Gates and Moore will save our butts :-)
That's not the
impression I get on web services-oriented and middleware-oriented
developers, but I don't
have enough first hand knowledge to judge. I would think that the prudent
thing is to explore
option Two, because I don't like Three at all, and Four has not appeared on
the horizon yet.
|
