Effect of using "native" jobs on boot time?

[upstart at eehouse.org]

> On Fri, 2007-10-19 at 12:33 +0200, Mildred wrote:
> > Personally I replaced init completely with upstart some months ago. And
> > I gained nothing in boot speed. Is was almost the same, or slower. I
> > can't say.
> > 
> Depending on what set of scripts you had before, and which you
> converted, this doesn't surprise me.  Simply converting scripts to
> Upstart jobs won't gain you anything, since there's only so fast you can
> fork() and exec() a series of monolithic tasks in sequence.
> 
> True speed improvements come when you *rethink* the boot sequence; start
> from scratch without referring to original scripts.  Identify what you
> actually need to do?

This makes sense to me, which is why I'm hoping Alex will have
something to say about frugalware.  Those distros I've seen using
upstart, including Ubuntu "Gutsy", simply replace rcS run out of
inittab with an rcS job: all the work is still done serially, with the
ordering determined by the names of links in e.g. /etc/rc2.d/.  But
Alex posted a series of jobs that truly did rethink the ordering in
terms of what required what before it could start.

I just launched the latest frugalware and note that it uses standard
sysvinit, and so I'm all the more curious whether anybody's seen a
speedup switching to upstart "native" (i.e. not rcS-based) init jobs.

Thanks,

--Eric
-- 
******************************************************************************
* From the desktop of: Eric House, xwords at eehouse.org                        *
*     Play one-handed with Crosswords 4.2 for PalmOS: xwords.sourceforge.net *
******************************************************************************