Re: Definition of _boot_
Excerpts from Ben Hutchings's message of Sun Apr 29 16:27:54 -0700 2012:
> On Sun, Apr 29, 2012 at 11:11:08PM +0200, Svante Signell wrote:
> > On Sun, 2012-04-29 at 21:52 +0100, Ben Hutchings wrote:
> > > On Sun, Apr 29, 2012 at 09:51:37PM +0200, Svante Signell wrote:
> > > > Hello,
> > > >
> > > > In line with the recent discussion, lets aim at defining what _boot_ is:
> > > [...]
> > >
> > > No, let's not. Beyond RAM, CPU, IRQ controllers and timers (all
> > > of which are part of the kernel's early initialisation) pretty
> > > much all of this varies from system to system and potentialy from
> > > boot to boot.
> > I'd assume that. Thank you.
> > > Further, if I normally log in to my laptop through gdm then gdm most
> > > certainly is part of the boot process *on that laptop*. And if I set
> > > up a Debian-based system as a web kiosk, starting the web browser is
> > > also part of the boot process *on that system*.
> > Then why can't we define what _boot_ is then? Single user, multi-user,
> > desktop, laptop, server, with or without X (soon wayland) etc?
> Ignoring special boot modes for the moment, the system is booted when
> its usual services are available. Exactly what those usual services
> are is determined by package selection and local configuration. We
> cannot make an exhaustive definition.
I personally distill it down to this:
A system is booted when it can perform its intended function.
I do not have a laptop so that it can start udev, or fsck its
filesystems. I have it so I can interact with it. So, when my endpoint
UI is available, it is booted.
I do not have servers so they can configure bonded interfaces, or start
an MTA. I have them so that they can serve web pages, or database
queries, or queue mail.
So, to me, to "boot" a system is to bring it into service completely.