I'm not sure that I like the idea of a 3D API being co-opted in an OS war here.  Truth is that is if > 90% of a target market are on the same platform then you don't actually need an open standards API at all; you can use something quite proprietary and still get a result.  That's real freedom - freedom to choose what works best for your project and what actually helps you to get stuff done without being constrained by artificial limitations such as "will it be portable?"

I likewise don't believe that desktop Unix is ever going to happen.  There's an awful habit of infighting that is deeply embedded in Unix culture, has bubbled over into open source communities (which seem to have become more about Unix than they are about open source) and results in a splintered, fragmented set of platforms, multiple forks, multiple different and incompatible ways of accomplishing the same end result, each striving towards their own particular vision of technical perfection (it's an interesting exercise to compare that with the evolution of GL extensions).  That's not conducive to widespread uptake; you spend more time trying to nail down a moving target than you spend actually getting productive work done.  It's ironic, and particularly telling, that the major success of Unix uptake (outside of specialist/server areas) has been in a locked-down environment where you get what you're given and have to learn to be happy with it.

I'm not sure that I like the idea of a 3D API being co-opted in an OS war here.  Truth is that is if > 90% of a target market are on the same platform then you don't actually need an open standards API at all; you can use something quite proprietary and still get a result.

I likewise don't believe that desktop Unix is ever going to happen.  There's an awful habit of infighting that is deeply embedded in Unix culture, has bubbled over into open source communities (which seem to have become more about Unix than they are about open source) and results in a splintered, fragmented set of platforms, multiple forks, multiple different and incompatible ways of accomplishing the same end result, each striving towards their own particular vision of technical perfection (it's an interesting exercise to compare that with the evolution of GL extensions).  That's not conducive to widespread uptake; you spend more time trying to nail down a moving target than you spend actually getting productive work done.  It's ironic, and particularly telling, that the major success of Unix uptake (outside of specialist/server areas) has been in a locked-down environment where you get what you're given and have to learn to be happy with it.