On Sun, Apr 09, 2000 at 04:54:22PM -0700, Gerald Gutierrez wrote:
> >My next point is maybe even more important, where are it all heading. What
> >is the goal of the project. To demolish bugs. Naahhh... Rather it should be
> >to create something new. But how and with what means. I don't know much
> >here... Actually I know nothing. But I think this issue is to or should be
> >solved soon.
>
> I feel the same way as well. The general consensus, as far as I can tell,
> is that it will end up being a part of a "better UNIX" (Ya or nay please).
> From my perspective I'm doubtful whether this goal will bring the HURD
> into any kind of spotlight or raise anyone's eyebrows. Its structure allows
> it to be optimized for any one of a number of applications, such as serving
> web pages or processing video, but I haven't felt any strong opinions. The
> way it's going right now, more than likely it will become another typical
> UNIX system foundation.
I don't know if there's much value in deciding an Ultimate Goal for the
Hurd (aside, possibly, from global domination <grin>) beyond what has
already been spec'd out for it. The plan to create a flexable
architecture where the user can even change the binary interpreter
is a hefty enough one as it is.
At full functionality, there's nothing stoping any sysadmin from using
the Hurd in *any* way they see fit, from RealTime applications to
security, etc. When the Hurd can reasonably be seen as the best answer
to the question at hand, it will gain in popularity - and I don't think
that will be far off. I have demonstrated loading a new pfinet while my
machine was under attack (Is there any other way to describe being
slashdotted?) and had the system still permit me to use the console
without much difficulty. This makes it already one of the choices for
what I would consider for a web platform. When we can do the same with
security, it becomes a serious contender as a campus fileserver that
isn't permitted downtime. When process migration between kernels becomes
real, it becomes a serious contender for clustering applications where
Beowolf is being used right now.
For those of us who can't kernel-hack, I beleive that there are 2 items
that need our attention:
1) Find bugs and build simple reproducable test cases. This is both
kernel-related, and application related where problems are specific to
the Hurd.
2) Write documentation to help others not have to get through the sweat
that we've gone through so far. I (and others) would cheerfully post
Hurd HOWTO's on web sites.
<Self-deprecating disclaimer reducing self to a mere 2 cents implied, but
not included>
--
There is no sin except stupidity.
- Oscar Wilde
