From: Paul <bd...@nt...> - 2004-06-23 22:32:28
|
Hi Kirill On Wednesday 23 June 2004 10:28 pm, Kirill 'Big K' Katsnelson wrote: > > A patch to fix a (perceived) problem in the kernel make system is not a > > path I would want to follow. > > ??? Why? This is just a small part of the vanilla kernel patch 2.6.5 to > 2.6.6. 2.6.6 and beyond would not require it since it is already > "applied" in a sense. I must admit I haven't been following the development of the 2.6 kernel - Been out of the loop for a while. > Does an RTLinux version exist for kernel 2.6? I was assuming it was > abandoned in 2000 or something - at least the version "freely" > "available" from fmslabs.com looks like it is. The last I heard on this front was the adoption of the adeos patch for 2.6 support. A quick look at the rtlinux-gpl cvs logs would indicate that progress is slow to non-existant on this front. However, I wouldn't discount the prospect of further development. > Anyway, patching the kernel build system feels rude to me, too. There > are also some technical non-problems with it, <snip> > What so you think about adopting it? The rationale here > is that it will give us uncomplicated and 2.6.6+ compatible makefiles, > so with the kernel 2.6.6, we can use its build system for modules only, > and with lesser kernels, would use the adopted one? I'm all for a simple build system. The majority of the sources are either user space or kernel code - Only a couple of files really need to be compiled for both (posemath.c springs to mind..). However, I have little interest or incentive in embarking on a new build structure for the emc/rcslib trees. I would rather devote time and effort to the emc2 tree where (hopefully) the layout is more conducive to a 2.6 build. Regards, Paul. -- "To err is human...to really f*** things up requires the root password." From a collection of quotes at http://www.indigo.org/quotes.html |