Protocol Specification

[Christopher Horvath]

Our current specification for protocols is ad-hoc, inconsistent,
incomplete, and messy.

Going forward, we're going to need the protocols to be an "official"
part of the specification. The file format is really only as good as
the protocols we can all agree on.

I would like to look at how OpenGL specifies its extensions and use
something like that. Sometimes I find the preamble in the extension
specifications a bit longish. I'd like the design discussion to come
after the formal specification, rather than (sometimes) before. But
otherwise, it's a good one.

Furthermore, I'd like the protocols we have now to all be tagged with
something indicating that they're the "old" set. So, much like GL has
this:

// Formally an extension
GL_EXT_point_sprite

// Approved by the architecture review board
GL_ARB_point_sprite

// Designed and supported by a particular vendor
GL_NV_point_sprite
GL_ARB_point_sprite
GL_APPLE_half_float

--------------------------

I'd like us to start of with taking our existing protocols, as they
are, and call them:

OLD_catmull-clark
OLD_polygon
OLD_particle

or something similar. We can then promote them to "official"
protocols, which should have some kind of prefix:

OGTO_particle

once they pass some kind of acceptance test. That, also, is a good
question - how should we "approve" protocols?