OpenGL ++ ARB Interest Subcomittee Meeting Notes
September 8, 1997
SGI progress report:
really working on the implementation and specification
[ Cosmo 3D/OpenGL++ differences (list same as last time) ]
Optimizer has MP support: queued updates w/transaction manager
Performer not flexible enough for Optimizer
SGI just starting Optimizer port
What we do have:
VRML subset (no routes, exotic nodes)
1/2 code, 2/3 functionality of Cosmo 3D; 1st cut at multiprocessing
We need a schedule for releasing code to subcommittee members
2-3 more months before alpha,
with a couple of spins on various parts
attempt at systematic testing of implemented functionality
How will conformance tests work? e.g. bounding box calculation: how tested?
Many tests are straightforward, though.
can compare performance to non-multibuffered case.
multiple contexts: not part of the MP framework
Need thread private data
may need to be fast access
[SGI's MP slides are not shown here.]
What is the exposed data structure?
decorated scene graph?
list of specific records?
For all shape nodes, need transform, appearance, environment
node. This info is sorted, then sent to the draw.
Review of other spec changes
Basic types cleaned up
Containers and fields are complete and match the implementation
Routes are gone (to be built on notification, which is still not well-
Much more complete viewer and window interface
MP still not present
Java bindings for OpenGL++
designed with Java in mind, but bindings won't be available for a while.
(mostly engineering resource issue)
PC version status:
Just barely up.
What about COM bindings? C bindings?
until OpenGL accepts compressed data, it doesn't belong in OpenGL++
- it is a file loader issue
- don't want to expose a compressed node
Do the OpenGL ARB by-laws need modifying to cover this? We need to
make sure the spec and other work are covered.
The group walked to Intel, where Feng treated us to an Intel
Scene Manager demo.