On Mon, 2013-02-11 at 13:56 +0100, Ingo Molnar wrote:
> To use another, perhaps more applicable analogy:
> 
> If one has the choice to start a new business in the U.S., it 
> would be reasonable to do that. There's a lot of supporting 
> infrastructure, trust, distribution, standards, enforcement 
> agencies and available workers.
> 
> Could the same business succeed in Somalia as well? Possibly - 
> if it's a bakery or something similarly fundamental. More 
> complex businesses would likely not thrive very well there.
> 
> *That* is how I think the current Linux kernel tooling landscape 
> looks like currently in a fair number of places: in many aspects 
> it's similar to Somalia - disjunct entities with not much 
> commonality or shared infrastructure.

That's complete nonsense. If you want to use pieces of the kernel
infrastructure, then just *take* them. There are loads of projects which
use the kernel config tools, for example. There's no need to be *in* the
kernel repo.

And for code-reuse it's even easy enough to automatically extract parts
of kernel code into a separate repository. See the ecos-jffs2 and
linux-headers trees, for example, which automatically tracked Linus'
tree with a certain transformation to make them sane for just pulling
into the relevant target repositories.

-- 
dwmw2