As pointed out by others, you could, but that's a pretty brittle
system.
Here are some random thoughts, not a proposal yet, but a place to start
discussion:
The URN mechanisms are not coming about as quickly as some of thought
they would be. Ideally, if I'm building a kitchen, and I want a teapot
on the stove, if I don't care too deeply what it looks like I should just
be able to WWWinline a "teapot.wrl", or something not much more complex
than that, and scale it with a bounding box. Where that teapot.wrl comes
from isn't *that* important (if it is, I'd specify an EXACT url)... if
the client hasn't heard of it, it can ask a couple close-by servers,
until it resolves it somehow.
Unfortunately, it looks like the URN mechanism won't be quite that flexible
any time soon. Which is understandable, as this is an extremely tall order -
beyond the technical problems in resolving URN's and setting up sort of name
authority, there are semantic and cultural issues. One man's teapot is
another man's chamber pot.
Okay, so let's say instead that a couple very prominent 3D
companies/institutions with fat pipes got together and created a very large
library of common, static, no-need-to-be-changed simple objects.
Collaboration would be crucial so that we don't have 10 teapots that are just
barely different. Now let's say these institutions made all these objects
publicly available as URLs, and provided some level of guarantee to their
existance - they didn't necessarily promise they'd always store the objects,
but they at least promise that the URL's will always work, if not directly
then by redirection or DNS intelligence. Once a common library is
established, let's say that library is put on a CD-ROM (thousands and
thousands of building blocks should be able to fit on one CD-ROM, and even
more with upcoming digital disc formats). Finally, browsers are modified so
their disk caches check this CD-ROM in addition to its regular disk caches
when accessing a given URL. In fact, if someone were to develop a
client-side proxy cache server for this purpose, no browser would have to be
modified.
So, to sum up, I might inline into my kitchen
"http://vrml-objects.sgi.com/kitchens/glassware/teapot.wrl", and feel
confident that people with a particular CD-ROM will have vast access to
that object.
The big problem I see is political - usually only one CD-ROM can be loaded in
a player at once, and being told to swap between CD-ROM's would Really Suck
(tm). So, the CD-ROM must have only the most essential objects, and keep the
fluff to a minimum. Perhaps it should be the domain of the W3O to issue
these... the nice thing is that this works perfectly well with today's
browsers (if client-side proxies are implemented) and WWW proxy/cache system,
and as the latter gets better this will still be useful.
Brian
--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@organic.com brian@hyperreal.com http://www.[hyperreal,organic].com/