| First I said:
|
| > 3) Desire for dissimilar contexts to co-exist without knowledge of any
| > other context.
|
| Then David said:
|
| > I believe this is a (n)on-goal. In order to allow multiple vendor
| > implementations of OpenGL to co-exist on the same platform,
| > a binary standard is required for dispatching that all dd code
| > and therefor contexts must be vaguely cognizant of. It needn't
| > be complicated - a simple name-based registration mechanism and
| > a jump table would suffice. This is nothing
| > more than part of the binary standard.
|
| Huh? How can this be a non-goal when you then assert the almost identical
| position and implementation suggestion I did? What we have here is a failure to
| communicate :-)
oops, I have an implementor's hat on - my failure. I think it is a
non-goal to make the implementation of a context unware of other
contexts and other devices in the system. I assert that in order to
provide interoperable vendor implementations there will need to
be a well defined binary standard at the dispatch level.
You are of course correct, that from an application programmer's
point of view there should be as little baggage in the way as
possible with respect to context differences. This was in
fact how we got to things like relaxing similarity constraints.
david
