From a Mesa perspective, Gallium3D is just a new HW driver layer at the lowest level of the Mesa code, alongside all the existing HW driver abstractions. You still need Mesa if you want to run OpenGL applications. Mesa is roughly a million lines of code; Gallium3D won't be replacing all of that.
We plan to stick with the old Mesa HW driver model until we have basic 3D functionality going for our latest GPUs (for the simple reason that developers know it and just finished bringing up the same functionality on 5xx parts), then to jump across to the Gallium3D-based code paths for all subsequent development. We think that will get functionality into our users hands most quickly -- and most of the code we write will be directly useable in the Gallium3D drivers.
One of the things I don't like of Mesa3D and such is that isn't community based development but done by a corporation. They are hiding some parts of the development phase like shamed of something dirty they are doing with the code, instead following an open and community based development process.
Ha! You're pretty funny dude. Mesa is a community-based project, and luckily enough, there are some corporations ( Intel, AMD, Tungsten / VMWare ) as major players in the project, without which not a lot would be happening. There is no requirement for anyone to instantly release any new code they come up with, minute-by-minute. What would you have them do ... all come around to your place and do all the development in your lounge room so you can report any dodgy happenings that you think someone should 'be ashamed of'?