[Syllable-kernel] Good artists borrow, great artists steal!

[Jake H. <jh...@po...>]

It was Picasso who coined the phrase in the subject of this message. 
One reason I enjoy developing for Syllable is that we're free to take 
high-quality tested implementations of things like glibc and GCC from 
GNU/Linux, thus saving developer effort that can go towards improving 
the Syllable-native code instead.

One question:  is there any reason, licensing or otherwise, for us not 
to also steal high-quality code from FreeBSD?  The BSD's already use 
GNU-licensed code and I believe the Linux kernel has used BSD-licensed 
code as well.  I'm not aware of any legal issues involved in intermixing 
the two, along with the GNU-licensed code written by Kurt, Arno, Daniel, 
Vanders, myself, and others, but I want to make sure no one else has any 
objections.

Here are the particular pieces of BSD that I'm interested in:

* kernel malloc():  we need a better implementation than the Linux 2.0.x 
malloc() we're using now.  Unfortunately, the Linux 2.2.x - 2.6.x 
"slabs" allocator uses a lot of global variables and is scattered across 
several files.  If the FreeBSD allocator is superior, then I would 
prefer to import that instead.

* FreeBSD NFS implementation (unless NetBSD or OpenBSD is better?). 
Linux's NFS has never been very good.  FreeBSD is beginning to support 
NFSv4, while Solaris 10 is freely available for interoperability 
testing.  Of course, for desktop users, Samba will be more important, 
however for interop w/ UNIX systems, or even Darwin/OSX, NFS support 
could not hurt.

* any relevant code from FreeBSD's TCP/IP stack, keeping intact our 
Syllable-native TCP/IP implementation while adding any new features or 
performance enhancing algorithms.

Anything else?

-Jake