On May 31, 2005, at 4:11 AM, Christos Zoulas wrote:

> | Well, the libc and libpthread developers thought they needed it.  I
> | assume any other developer writing assembly code might find it
> | useful, too.
>
> These are much more attached to the kernel than regular application
> developers.

I think that's nonsense.  Just because parts of these libraries are  
tightly-coupled to the kernel doesn't mean that their use of genassym  
is.  These libraries have assembly code in them that accesses data  
structures.  How would that be different than, say, a hand-optimized  
video codec or a cryptographic transform?

> This is still the case. From where I am sitting the number of non- 
> kernel
> developers far outweighs the number of kernel developers. In anycase,
> if we move them, we should move them both.

Your use of the term "outweighs" suggests that moving genassym would  
be somehow harmful or against the interests of non-kernel  
developers.  "Outnumbers" would be a more appropriate term, I think.

In any case, I don't think all of the embedded system developers out  
there are system administrators.  And they configure kernels all the  
time.

-- thorpej