I agree that for some tasks it makes sense to let
the tool do the optimizing
but this depends on what the issue is.
We still compile code into machine code in the end because low level
offers
more power and versatility
than high level code.
For high level GUI
editors you leave more possibilities open if the format is
powerful
and therefore low level
formats are more likely to be welcomed by both newbies and
advanced users in the long
term unless perhaps the newbie wants to hand code and
this too would depend on
how the format has been designed.
----- Original Message -----
Sent: Thursday, December 21, 2006 1:38
AM
Subject: Re: [x3d-public] dedicated
working-group focus on X3D interoperability
Don;
Congratulations on this working group, and of the moratorium on new
nodes. Most certainly, the number of nodes in X3D, and the rate
they
are being placed, is staggering.
Can I email my thoughts on this?
Athough FreeWRL tries to support as many nodes as possible, I think
that the X3D "problem" could be solved differently. This view comes
from many years of hacking about with FreeWRL.
Quite simply put, all of the Triangle and *2D nodes could be
dispensed
with. Same with some others. Focus should be on optimization
techniques
to render shapes quickly, on given hardware.
What we appear to be doing is to move that optimization level up the
food
chain; making our job easier. That's not correct - we should be
making
the job of the authors (and, authoring programs) easier.
X3D has missed the mark of abstracting the details of the
graphics
implementations from the users. It should be up to the browser
implementer
as to how to best send the data to particular hardware.
If one can infer a relationship with computing languages, We
are producing
a "super assembler" - one that can do anything. The world, except for
certain
real time programming, does not use assemblers any more; it's much
easier
to write in Java or C or whatever, even if it is maybe not as
efficient. When we
write in Java or C, we expect the compiler/interpreter to make
optimizations for
us *that fit with the current hardware*.
X3D seems to have missed that point.
FreeWRL, by the way, treats most geometry identical internally. The
end goal
is to perform runtime optimizations on the data; whether that data
comes
from an Extrusion, ElevationGrid, IFS, TriangleSet, Mesh, whatever. We
are
not quite at performing major runtime optimizations yet; but it will
come.
Anyway, my two cents (Canadian :-); If you, or anyone else on the
recipients
list wishes to discuss this further, please email.
-----------------------------------------------------------
John A. Stewart
Team Leader: Networked Virtual Reality.
Network Systems and Technologies -
Systemes et technologies des
reseaux
Communications Research Centre Canada |
Centre de recherches sur les
communications Canada
3701 Carling Ave. | 3701, avenue Carling
PO Box 11490, Station H | CP 11490, succursale
H
Ottawa ON K2H
8S2 | Ottawa (Ontario) K2H 8S2