>  An important maxim in software engineering is that "correctness is a 
>  constraint, while performance is a goal." The simplicity of the new 
>  threads implementation makes it easier to deliver correctness, and 
>  allows more time to be devoted to performance tuning.

I was under the impression that we would be saving the kernel a lot of 
trouble by using userland threads whenever possible. and thats obvious 
because the kernel doesn't have to deal with the overhead of that thread. 
how do you go about denying that?
 OTOH, the implementation of threads in itself makes v little difference 
either in (terms of) functionality or performance. so maybe a 1:1 is 
really the best thing to have.
regards
-kamal