On Thu, May 15, 2008 at 12:01:22PM -0700, [EMAIL PROTECTED] wrote:
Such a shame. I've always wondered if RMS would have more glory today
if Minix or a BSD were available to him in 80s with an acceptable license.
Keep in mind that RMS didn't work on the HURD. As far as I know, he's
never edited a single line of the code. Michael Bushnell did most of the
initial work. Roland McGrath helped later, and there are a handfull of
other people involved.
I think the main reason that the Linux kernel hasn't forked is that Linus
figured out how to do distributed development. It really isn't as much
that the Linux kernel doesn't fork, it's that it has numerous forks, and it
isn't a problem, because people can work with them. There are many Linux
trees out there, and this is generally considered a good thing.
Even after the AT&T mess was done, seems he could have *still* taken the BSD
source, wrapped it with his GNU tools and called *that* the GNU system.
Oh I don't know, you see how successful he's been in getting it called
GNU/Linux.
Hell, he could fork FreeBSD and do that *today*. Then we'd finally have
a polished GNU system.
No real point. They already officially support two obscure Linux
distributes, since those distributions refuse to allow proprietary code to
be included.
L4 definitely takes a different approach. I think Mach didn't end up being
a micro-enough of a micro-kernel.
Being micro is the big draw of a microkernel to me. L4 looks appealing now.
L4 removes a lot of what was even in Mach. I guess the HURD developers
decided that L4 wasn't sufficient for their design (it doesn't have
capabilities).
People are doing some neat things with L4, especially with efficient
implementations, showing ways that a microkernel architecture can be more
efficient than a monokernel. Starting with Linux on top seems like a good
way to migrate toward the architecture.
David
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
