Supporting a GNU Hurd port?
john.r.moser at gmail.com
Wed Dec 9 15:07:44 UTC 2009
On Tue, Dec 8, 2009 at 12:22 PM, Lefty (石鏡 ) <lefty at shugendo.org> wrote:
> On 12/7/09 10:38 PM, "Danny Piccirillo" <danny.piccirillo at ubuntu.com> wrote:
>> I think it would be cool, but are there any reasons against this?
> 2) the number of people who would derive even the slightest bit of
> benefit from it number, perhaps, in the dozens.
I object to this reasoning only on the grounds that you have
absolutely no way of actually validating this statment or any basic
theory that lead you to this belief.
Moving to a microkernel could, in theory, produce a massive market
paradigm shift. Moving to a real, engineered-from-scratch
microkernel, however, could lead us to a new golden age of computing
blahblahblah, you know the microkernel arguments, and they're actually
pretty considerable. The idea of a system that's easier to maintain
(face it, operating systems are huge now; smaller chunks are easier)
and self-error-correcting (MULTICS did it, Minix does it) is not
really far fetched, not any more than building a machine that can fly
or building a device that produces light without burning something.
(ASIDE: Linux is "going towards a microkernel architecture" so they
say, but really it's just a kludge with a bunch of crap moved out of
the kernel; a REAL microkernel is a completely different project, and
even Linus would agree that if he wanted to go that direction then
it'd be time to step completely off the Linux code base and start from
On the other hand, moving to a microkernel could, in theory, produce a
lot of headaches, a lot of wasted time, and confirmation that the
above is simply BS. The typical microkernel anti-arguments (it's
slow, some major drivers have to be moved inward, etc) could just be
shown absolutely true in the state of the art. Some small scale
stress testing and benchmarks show Minix is actually better and
capable of working as advertised; but it's never been deployed as the
heart of a real, desktop-ready system like Ubuntu, and THAT benchmark
is completely different.
I submit that we're as a whole not in the least bit qualified to make
definite statements about the benefits of moving to a new kernel
architecture. This is a chicken-and-egg problem, of course: if you
want to discuss this seriously, then someone has to put in the
engineering time to get Ubuntu barely functional on top of i.e. Minix
or Hurd, and test it out; but we don't know ahead of time if that
engineering time is worth spending in that way. I am, by the way, a
Minix fan, not an L4 fan.
(by the way, the above chicken-and-egg problem is identical to many,
many social problems, especially in the context of personal
relationships. If you can't understand it as an engineer, then
understand it in that context: how do you convince someone it's worth
leaving their current relationship partner for a new one, when their
current relationship isn't particularly distressed and the new person
isn't particularly remarkable at face value?)
More information about the Ubuntu-devel-discuss