In making A3DL we did take some inspiration from VRML.  However, VRML
(and X3D) have some aspects that if adopted directly would make A3DL
a bit less efficient or more cumbersome.  We have all the structure that
comes from VOS to use.  X3D does some stuff just because VRML did it,
and VRML97 does it because VRML 1.0 did it, and VRML 1.0 did it because
Inventor did it, and SGI GL,  etc. etc.  

We should do what is closest to overall common practice. i.e. if VRML,
CrystalSpace, other 3D engines, Collada, etc. all do somtehing in a very
similar way, we should probably do that.   In some cases we will have to
break with X3D (or other existing stuff) though and do it better (better
for VOS).

The two overarching design goals for A3DL are to first, make it work
efficiently and be flexible and useful to manipulate A3DL objects in and
of itself.   Second, to make it easy to import existing content.  These
two things will be in conflict sometimes.  I don't know which should be
primary, but I lean towards making A3DL itself easy to use (e.g. imagine
you're interactively editing A3DL objects with a GUI version of mesh)
and very flexible (e.g. imagine that you want to share content between a
3D and a 2D application, or between 3D worlds, you want to make something 
clever and neat in A3DL.)

(Also, in the "designing metaobject structures" vein, there's a Wiki
page about "Tips and Best Practices" here:
http://interreality.org/wiki/TipsAndBestPractices . I started it out
with a few tips and TODO items, add more TODOs (and tips) as you
encounter situations and figure things out.)



Reed

_______________________________________________
vos-d mailing list
vos-d@interreality.org
http://www.interreality.org/cgi-bin/mailman/listinfo/vos-d

Reply via email to