[CVALE] Linux Kernel Developer Quit

[Dennis Baker]

> Why bother modularizing then?

So when I boot my desktop I'm not loading the AMI Megaraid driver or the XXX
fiberchannel controller drivers.

Driver support for Linux is great, other than the few proprietary drivers
out there or cutting edge stuff it's awesome.  I actually find most drivers
under Linux work better than their Windows equivalents (and have for some
time).

As for ReiserFS requiring a recompile of the kernel... what other OS
supports multiple file systems the way Linux does?  I've lost track but
Linux ships with support for 6 or 7 different file systems... Reiser could
be in there but they chose to do their own thing.  As Pete said they could
also use the standard interfaces but they want to add functionality which
goes beyond the standard.

-- Dennis

On 7/26/07, Pandora <cvale2 at synx.us.to> wrote:
>
> Landon Blake wrote:
> > http://www.nuxified.org/article/fork_a_kernel_kill_an_os_and_revolutioni
> > ze_the_desktop
>
> fixed ->
>
> http://www.nuxified.org/article/fork_a_kernel_kill_an_os_and_revolutionize_the_desktop
>
> > Any thoughts on the recent exit of a kernel developer?
>
> Eh. More room for me! Doesn't seem a good reason to fork though.
> Desktops and servers are both very similar in function, and I don't
> really think it's a good idea to encourage strong separation between the
> two, because then you end up with a few elite servers making a mint off
> their embedded advertising, and everyone else is a desktop user, unable
> or unwilling to fight past the barrier you have created to become
> servers themselves. Why can't we all just be peers?
>
> My complaint about the Linux kernel is, it needs to be more modular. I
> find it terribly ironic that they added very good support for 3rd party
> modules into the kernel, but then they still package 99% of those
> modules in the kernel source distribution. Why bother modularizing then?
> We should be able to move all the modules for all the hardware out of
> the kernel, and have it still compile, then have all those modules
> compile from outside of the kernel source tree, and still be able to run
> a working system. Imagine how small the kernel would be, without any
> hardware support!
>
> Also kernel patches really annoy me. Reiserfs should have been a
> standalone module, but instead it requires you run a patch that modifies
> your kernel sources any way it wants. That's not good modular design in
> my opinion. Plus it's damned hard to upgrade, since the strong coupling
> means when kernel developers change a file that has little to do with
> reiserfs, it breaks the patch which modifies that file.
> Documentation/Changes, anyone?
>
>
> I agree with this guy:
>
> http://www.nuxified.org/article/fork_a_kernel_kill_an_os_and_revolutionize_the_desktop#comment-9699
>
> _______________________________________________
> cvale mailing list
> cvale at lists.fire2wire.com
> http://lists.fire2wire.com/mailman/listinfo.cgi/cvale
>



-- 
Dennis Da-Ogre http://ogrehut.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.fire2wire.com/pipermail/cvale/attachments/20070726/abe61574/attachment.html