How about splitting rc.sysinit into multiple commands in a directory, init.d style? I've seen several cases when various applications (usually stuff that wants to do low-level things) needed to somehow insert themselves in the rc.sysinit sequence. With a monolithic rc.sysinit that's impossible unless you hack the script using sed, perl, etc., which is ugly to say the least.
One issue with splitting it (among many other that i'm not going to mention) might be that the boot up time might get a bit bigger. If i'm not mistaken, this could be avoided by creating one single actual executable, and let it source things up from a directory, instead of launching multiple executable scripts sequentially.
Splitting this executable might also allow for things like optionally turning on (or NOT turning on) IEEE1394 or USB in userspace (as opposed to using kernel parameters in the bootloader), etc.
Keyword: flexibility Downside: complexity
How about splitting rc.sysinit into multiple commands in a directory, init.d style?
It's been seriously considered, just not enough round tuits yet.
What's the policy on someone from the "civilian" world coming up with a solution? Will RedHat accept stuff like that?
-Chuck
On Sat, 2003-07-26 at 12:07, Chuck Wolber wrote:
How about splitting rc.sysinit into multiple commands in a directory, init.d style?
It's been seriously considered, just not enough round tuits yet.
What's the policy on someone from the "civilian" world coming up with a solution? Will RedHat accept stuff like that?
I think this is the key reason that the RHLP exists. We're opening the playing field here. We want everyone to be involved. If you can code, hack. If you can write, document. If you can't, bend, break, and file bugs. You don't have to end in @redhat.com to play.
If your code is good, it will be considered. Thats not to say that we'll be including everything, it has to be maintainable.
But if you can speed up rc.sysinit in a clean way by splitting it up, by all means, do it.
~spot --- Tom "spot" Callaway <tcallawa(a)redhat*com> SAIR LCA, RHCE Red Hat Enterprise Architect :: http://www.redhat.com Project Leader for Aurora Sparc Linux :: http://auroralinux.org GPG: D786 8B22 D9DB 1F8B 4AB7 448E 3C5E 99AD 9305 4260
The words and opinions reflected in this message do not necessarily reflect those of my employer, Red Hat, and belong solely to me.
"Immature poets borrow, mature poets steal." --- T. S. Eliot
On Sat, Jul 26, 2003 at 05:38:48PM -0500, Tom 'spot' Callaway wrote:
But if you can speed up rc.sysinit in a clean way by splitting it up, by all means, do it.
Personally, I think the actions in rc.sysinit fit pretty tight together and it will be pretty difficult to split up in a decent way. I could imagine that it hands off some tasks that *might* people want to customize to separate scripts, though.
Chuck Wolber (chuckw@quantumlinux.com) said:
It's been seriously considered, just not enough round tuits yet.
What's the policy on someone from the "civilian" world coming up with a solution? Will RedHat accept stuff like that?
Basic requirements:
- have an rcS.d for sysinit stuff - get all the ordering and requirements right - fix chkconfig, ntsysv, redhat-config-services, etc. to DTRT and present a sane UI to it.
Bill