[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
- From: "Bullard, Claude L (Len)" <email@example.com>
- To: Justin Couch <firstname.lastname@example.org>
- Date: Fri, 30 Mar 2001 09:36:35 -0600
Yet it sat there in use for the last five years.
Yes, the incompatibilities in the implementations
were glaring, the vendors rivalries enormous,
all true. Yet it sat there in use for five
years and the resistance to change has been
Partly a troll, but partly fact. The pace
of mucking with a spec makes it just as unendurable
as the work left undone. I also recall the
end of the VRML project (much like XML) where a
minimal victory was declared and everyone was
told to go and implement the left-over
parts away. They tried that. The event models
are a problem but the simple obvious bits were
the incompatible color models. Interaction
is important but in a rendering app, the
consistency of appearance is critical. So
people hammered on that model. Meanwhile, the
vendors began to die off. Experts went walkabout.
VRMLNextGen became a two encoding project.
We had a language in which the original object
model was tied directly to the syntax of the
file format. People rose up to defend the brackets
when they should have been looking at the object
model. XML was declared to be the death knell
because "everyone" knew a document model couldn't
be used for the object model. Few stopped to
look at the infoSet and some blithely accepted
that two encodings couldn't be harmful. They
locked horns on the syntax. Experts told them,
uhh, the object model first, brackets later.
But hey, the W3C wanted pointy so pointy it was.
You are right about the lack of understanding
for many of us. It was a learning experience.
But I also remember being told that HTML was
the "shining moment of clarity" and all of that
from those who designed the first object model.
XML, "the evil from the east" and so on. IOW,
stooges on both sides of the aisle. nyuk-nyuk-nyuk.
Competence: how high can you toss the balls when
Things that have multiple standard parents
rarely stay simple or on track. The only thing
that kept VRML stable was the five year cycle for
changes while all of the OTHER specs not yet standards
got hacked into some semblance of stability.
On this point you are wrong: the VRML spec group
was never closed. It has always operated in the
open. Some people left pissed off, that is true,
and some returned for hire or for entertainment, but
they never closed the doors. It is the one promise
kept. Now, did they keep on remaking the same
decisions? no. At certain points, they picked a
direction with or without understanding it and
went forward. XML was a bear. The DTD wasn't
up to the modeling challenges, Schemas weren't
done, XPath wasn't done. The DOM event model
wasn't done. MPEG was trying to move it into their
models complete with patented tech, some
wanted to replumb XML (just a DOM
away from being gay; SMLs funny), and on and
on and on, collision after liaison, expert after
standard. And so it went.
Meanwhile, VRML97 sat there fat dumb and happy
being used for five years. God bless ISO.
Interoperability? Heck, we'd just like to have
the same capability as the original VRML97 still
working, so we rely on the two vendors still standing
extensions and all. They are after all, experts.
And they both use VRML97: fat, dumb, and happy.
We can't have cooperation, negotiation, interoperation,
webbiness and so on without a certain reserve about
what is possible. Some call that minimalism, some
just do the obvious bits and don't need a name or a
religion for common sense. OTOH, it is almost
certain that the advanced bits won't get done or
won't be interoperable when done.
Complexity bites. So is simplicity the solution?
Sort of. It comes with its own ticket. The truth about the web
is that to have interoperability at scale, one becomes content with a
certain mediocrity in the applications and incredibly
wary of those that proffer simplistic solutions
for what are known complex problems. Daring to
do less means being able to do less. Trying to
do more with less often means taking profit
and turning it into customer bribes. When that
runs out, the customer becomes the patsy.
It becomes drug dealing in the webHood.
Last year, web businesses were giving away web services.
This year, MP3.COM charges artists for the privilege of being
paid for MP3.COM to use their songs, AT&T is charging
for being paid for its services (that's right; we
pay them to bill us unless we allow them direct access
to our bank accounts and give up the ability to inspect
bill prior to payment), and submarine patents are
being welded to open technology.
Why does Microsoft dominate? They don't care
unless it ships a million copies. They throw away
the rest and if you count on an app on the scrap,
you burn with the leftover DNA.
It is the cost of lowering the cost.
Scale is a key to the record business and the software
business, but not the scale of application,
the scale of sales. Farmers need large orchards
to make minimal profits off seasonal harvests.
Otherwise, fruit costs what it costs to grow
without migrant labor and heavy pesticides.
Organic tastes like industrial; it just rots faster.
Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h