Debian alternative kernels [OT]
consul tores wrote:
Yes, Linux (kernel) is very tweakable, but normal users are not able
to compile their own kernel; i am more remembering when i could
install using 3 diskettes, and now i can not do it anymore.
If, we consider that the environment has changed; we have Red Hut,
Ubuntu and Suse; pushing to include every thing into the kernel, what
is the best for them, then we have a huge kernel; which is not the
best for older ordenators, but it is the best for newer boxes. As we
can see, Linus is been pushed to built a huger kernel.
I'm sorry but this is quite wrong. Nobody's forcing anyone; on the
contrary, kernel developers *want* to integrate many things in mainline, and
for very good reasons. See this file[1] for some (about the driver model).
1: linux/Documentation/stable_api_nonsense.txt
See it's not the big distributions pushing stuff in the kernel, it's mainly
(among other things) new drivers to support more hardware, including very
old machines. Everything is extremely well organized and modularized, and
you're able to build a kernel with just what you need. Again, if you can't,
someone else can, that's what distributions are for; but you can't expect a
*general-purpose* distribution to strip off their kernels and drop support
for a range of hardware for no reason at all. As Stefan said, Linux does a
very very good job on embedded systems, you can't deny that.
If, Debian has a very tested own kernel (Hurd), it should be focused
to its users, who probably are using older hardware, and maybe are not
using non-free software. This is why, i think that having a Debian
kernel, the users could be covered against global decisions.
The Hurd is not Debian, it's GNU. Again, there's no "global decisions" at
the kernel-level (at least not about what you're referring to), the
distributions make the decisions of how they want to distribute the kernel;
if you're not happy with the Debian kernels, well, maybe you ought to
search for a more specific distribution (Debian itself has enough
derivatives). Don't forget we're talking on the -user mailing list of one
of the most universal projects.
The thing is, the Hurd won't change that. If it seems tinier, well it's
probably because it currently has less complete support. Microkernels don't
exactly require less code "in general" AFAIK, it's a matter of architecture
and where and how the code runs. I don't believe we should even think about
having the monolithic vs micro kernel discussion here by the way; both
Linux and the Hurd are very cleanly written and do what they can the best
they can, given their respective design limitations.
[snip]
No, not the development model; i am refering to the structure, a
monolitic base system, which is very small and stable.
Well, it's usually seen as such because it's tied with its core userland.
Debian uses GNU instead, so *in that regard*, I don't believe Linux and the
FreeBSD kernel alone are *that* different. I would rather compare them by
their development models, philosophies or simply licenses, for example.
Someone might be able to troutslap me one that one.
Yes, i think in the same way, we need to test Hurd in an efective way.
it could help to manage the actual tendency to emulate Windows,
obtaning a sipler/efective/funtional OS. I could be wrong, but it
seems the most of us are prefering stability.
A tendency to emulate Windows at the kernel level? I don't think so. This
matter is a matter of the userland; and there are many, many alternatives
to GNOME and KDE already - heck, even living without a prebuilt and
integrated DE isn't so hard. Anyway, testing the Hurd is certainly a noble
thing to do, but even if it gains traction, it won't get a mysterious
anti-WIMP DE that Linux already has. It's other needs it answers to.
-t
Reply to: