Hi,
On Sunday, September 29, 2002, at 08:13 PM, Matt Brandt wrote:
I guess I would lean toward something halfway between a bunch of
reader/writer classes and a database. It seems to me like the scene
graph really is the database, but we might need to add a little bit
of information to make things more explicit. Named nodes would be the
main thing I can think of, so that a scene object could be reused
without a great deal of work. If we try to get into the idea of
merging nodes I think we can end up breaking things inadvertently. In
some scenes an object might seem like a static object most of the
time and just need to be dynamic for one thing. Only the creator will
know for sure. Optimizations like that can be left up to the program
that creates the scene (or the person).
I agree that a database should mainly be used to serve as a clean
import/export interface. But I can imagine some sort of optimisations
which can be done on that level, ie. when optimising for a specific
target rendering device.
What do you mean by 'named nodes'? You already can name G3DGraphNodes,
can't you?