19 Jan
2012
19 Jan
'12
2:38 a.m.
There's not much justification from a technical perspective for the
behavior of the kernel maintainers in the CK patchset incident. One
of the grad students that worked with my group did profiling of Linux
kernel scheduling behaviors and he examined some of the work that
happened. What it boiled down to was the CK patches were rejected by
the guy responsible for the scheduler in Linux, said guy went and
wrote his own implementation of the scheduling strategy from the CK
patches which ended up doing a worse job than the CK patches, and that
implementation was the one that got merged into the mainline. He
found a whole lot of other crap in the scheduling behavior that were
problematic as well but that's a different story entirely. There are
plenty of cases in the Linux development world wherein ideology,
politics, and ego have stood in the way of technical advancement or
merit. By and large, the Linux kernel maintainers are not really
developing or maintaining Linux for use by others, they are doing it
for their own personal needs/desires. As far as I know, the kernel
maintainers still resist the very idea of a pluggable scheduler
architecture despite the advantages it would bring for various
end-users because THEY themselves do not consider it to be necessary.
The source code may be there, but if you can't get the changes you
need into the mainline, then the cost of maintaining the feature you
need approaches the unfeasible, at which point availability of source
code is pretty much irrelevant.
On Wed, Jan 18, 2012 at 6:21 PM, Adam <geekdundee(a)gmail.com> wrote:
Ah, politics. ;)
Linus Torvalds wouldn't even let a kernel debugger make it into the
kernel for a long time because he seemed to be under the impression
that they are for sissies.
http://linuxmafia.com/faq/Kernel/linus-im-a-bastard-speech.html
http://www.mail-archive.com/linux-kernel@vger.kernel.org/msg01462.html
(note: currently blacked out to try and rais awareness that some
shitheads in the US Congress want to censor the internet [with SOPA and
PIPA legislation] - try after ~12 hours)
(that being said the reply by Ameen Ross may have merit - there are
probably two sides to the story - the LKML should be able to tell)
On Wed, 18 Jan 2012 19:18:48 +0100
Timo Kreuzer <timo.kreuzer(a)web.de> wrote:
Am 18.01.2012 16:06, schrieb Aleksey Bragin:
_______________________________________________
Ros-dev mailing list
Ros-dev(a)reactos.org
http://www.reactos.org/mailman/listinfo/ros-dev
P.S. Proper answer from Linux team would be "We accept
patches!" :-).
Maybe that doesn't match reality. I remember reading an
article by
some guy who wrote lots of patches for the scheduler and finally
rewrote the whole thing.
It proved to perform better than the original scheduler and many
people were using it, yet it never got incorporated into the mainline
kernel. At some point some other guy rewrote the original scheduler
doing basically the same thing that he had developed earlier and at
that point he quit linux development.
_______________________________________________
Ros-dev mailing list
Ros-dev(a)reactos.org
http://www.reactos.org/mailman/listinfo/ros-dev