Re: Definition of _boot_
On Tue, May 01, 2012 at 10:36:58PM +0200, Svante Signell wrote:
> On Mon, 2012-04-30 at 20:30 +0200, Vincent Bernat wrote:
> > OoO En ce doux début de matinée du lundi 30 avril 2012, vers 08:15,
> > Svante Signell <svante.signell@telia.com> disait :
> >
> > >> I'm rather sure that he wants to define booting as part of what
> > >> currently is done in /etc/rcS.d. Configuring the network or mounting
> > >> non-essential remote file systems wouldn't be part of this definition.
> > >>
> > >> Then he would state that these early tasks do not need events at all,
> > >> and conclude that later tasks can be handled in event based userspace
> > >> tools, but that the initial process that invokes these event based tools
> > >> doesn't require events and thus can stay simple.
> >
> > > Nice summary, thanks. This is the whole idea behind defining boot...
> > > Some people get it, others don't.
> >
> > Since your boot definition is mostly the current initrd, you seem to
> > agree that the current init system could be replaced with something more
> > current like upstart and systemd.
>
> On the contrary, with this definition init scripts are sufficient, and
> the event-based stuff happens later, e.g. with event based user space
> tools, udev<->linux kernel.
So your definition is circular: booting is what you can do with the
current init scripts, therefore the current init scripts are fine for
booting. Unsurprisingly, you didn't persuade others to agree with
this definition.
Ben.
--
Ben Hutchings
We get into the habit of living before acquiring the habit of thinking.
- Albert Camus
Reply to: