Florin Herinean wrote:
Isn't it possible to perform a two step vrml parsing ? One to lookup what
actual nodes are used and establish the parsing strategy for the second
step. That way you can optimise the scenegraph, especialy for the two cases:
a) static scenegraph (nothing ever change or move) and b) unmodifiable
scenegraph (object may move, but no objects are added or removed).
We have a construct called StaticGroup which allows this type of
optimization. But in general you cannot analyze VRML content this way.
External scripting can be added at any time which expects to be able
to add/delete nodes. A caveat with StaticGroup, it works but isn't
heavily tested yet, I think it has issues with bindable nodes like fog,
background, viewpoint
In a strict loader case, ie no VRML runtime we could optimize this way.
We have discussed providing one of these, but our main focus has been
in getting a compliant VRML97/X3D system going. My sense is its about a
2 week task to break off a loader which would be optimized for j3d
loader usage instead of full X3D event model. Absent someone else doing
this work or us getting funded for it, it will likely be a low priority
for us.
--
Alan Hudson
President: Yumetech, Inc. http://www.yumetech.com/
Web3D Open Source Chair http://www.web3d.org/TaskGroups/source/
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
