[
Lists Home |
Date Index |
Thread Index
]
On Tue, 26 Oct 2004 14:21:48 -0500, Bullard, Claude L (Len)
<len.bullard@intergraph.com> wrote:
I'm five years older, possibly 5 years wiser, and definitely much
less convinced that I know what's best for XML or its users than I was
in 1999 :-) I don't have an agenda here, just a bunch of questions. I
honestly have no idea whether it's better to try to "fix" the things
people are complaining about or just build better tools and document
best practices to keep people from being snagged on them. I just find
it intriguing that after five years of wrestling with this stuff,
many smart people are ending up roughly where the SML wars started in
the fall of '99.
>
> The other approach would be that critical individuals agree
> (eg. Derek-Denny Brown, Dare Obasanjo, etc. at Microsoft,
> Tim Bray at Sun, whoever their peers are at Oracle, etc.) who
> then make the W3C aware that they will be moving to a new
> consensus. (I'm not discounting the W3C, just that this
> goes faster and more directly if the requirements originate
> on the vendor side.)
The vendors aren't going to do anything to reduce interoperability and
"standards" compliance, and the W3C won't do anything without a very
large push from the vendors. We're in an odd situation where
everybody (well, more or less, AFAIK) knows that there is a problem
but nobody has much of an incentive to do anything about it. It will
take some external shock -- such as what gasoline prices did to the
American auto industry in the '70's, or what Microsoft did to IBM in
the 1980's, or what the Web did to Microsoft's desktop plans in 1995
-- to shake things up, IMHO.
It's possible that the shock will come from customers who realize that
the whole XML corpus is massive overkill for their needs and they
shake things up by favoring the products/projects that have
anticipated this and offer highly optimized price/performance for the
core use cases. Or the mutant things descended from standard XML that
find a niche in wireless apps or industrial-strength message
processing may spread outward. Or a severe economic downturn could
make bandwidth and processor capacity scarce commodities again. Or
maybe end users exposed to XML in Office or whatever latch onto its
potential, but rebel at the cruft. I really don't know, but sooner or
later the days of huge specs that few really understand and can only
be implemented "properly" by massive interop testing will come to an
end.
What I think would be useful in the short term is to:
- Better understand which parts of XML, the related specs, and the XML
APIs are really being used.
- Use that knowledge to optimize products and profile specs for the
most common use cases, even if there is no real willingness to change
the specs.
- Let those who need to use the subsets, or out-of-the-mainstream
ideas such as alternative infoset serializations, do so in their own
little universes. Wireless comes to mind: they need much of what XML
offers, but the whole package is just too much.
- When shock comes, we'll have experience on which to base
standardization choices, much like the XML WG had when the Web shock
hit SGML.
> I expect that some parts of this question are answered
> by your presentation in DC next month, yes? If so,
> we can take it up in the hallway.
I wish I had answers! I did end the formal paper with something along
the lines of "The main question in my mind is whether the pains people
are getting from XML will be addressed by the W3C in incremental
revisions, or by something dramatically different and incompatible
that comes in from the outside. I prefer the former, but predict the
latter."
|
