I took a look at linux streams implementation and that is completely
out-performing. We could do really better if there was critical mass
enough to start this project.
Not to be a skeptic (OK, to be a skeptic) what solid data supports
this linux streams thing being "completely out-performing"? Not to
bash the Linux folks, but their enthusiasm and exhuberance many times
overwhelms their ability to judge accurately.
Probably it was worked on by Jim Freeman at Caldera, who did a lot of
work on the Streams frameworks in UnixWare (SVR4.2 SMP), AIX, and
Solaris before he left Novell/USG, and seriously knows what the hell
he is doing.
On the other hand, it could be some other guys, and I could be wrong. 8-).