Re: .Net/Mono and C# in GNOME? Do we really need it?

[Peter Dolding <oiaohm gmail com>]

On 12/29/06, Alex Jones <alex weej com> wrote:

> >
> > On Sat, 2006-12-30 at 10:06 +1000, Peter Dolding wrote:
>   
> > > > Alex Jones wrote:
> >     
> > > > > > On Fri, 2006-12-29 at 13:02 +1000, Peter Dolding wrote:
> > > > > >
> > >       
> > > > > > > >>> "initrd" is a RAM Disk used for initialising a bootup process. I have no
> > > > > > > >>> idea what relevance that has.
> > > > > > > >>>
> > > > > > > >>> If you mean SysVInit, well... /it's already gone in Ubuntu/. Upstart
> > > > > > > >>> replaced it ages ago.
> > > > > > > >>>
> > > > > > > >>>
> > > > > > > >>>
> > > > >           
> > > > > > >> PS the first SysVInit to run is inside the initrd.  The mess starts at
> > > > > > >> initrd and moves all the way threw the boot process.
> > > > > > >>
> > > >         
> > > > > >
> > > > > > SysvInit isn't run at all. It is already replaced by Upstart.
> > > > > >
> > >       
> > > > The initilization min SysvInit last time I looked is still there.  That
> > > > just starts what drivers and setup is required from the initrd is still
> > > > there before starting /sbin/init.
> >     
> >
> > Are you sure you're not confusing SysVInit for the Linux kernel starting
> > up?
> >
No.  I have edited inside the initrd mainly because of moving machines 
between hardware and the initrd being wrong for the new hardware and 
setup..   Lot of distros start a small form of init from started with 
linuxrc inside initrd.   I am not talking about replacing the kernels.  
I am talking about fixing up the initrd startup system from being copied 
from SysvInit and requiring ash or nash or some form of sh to operate 
inside the initrd. To something hopefully smaller and faster.

5 years ago to be more correct 6 years ago the idea of replacing the 
SysVint started as part of Linux Standard Base.   Many have came and 
gone most have missed that its so far embed its not funny.

Problems some distro's when puting stuff into initrd just copy the 
normally SysvInit files in and use a min SysvInit to start them.  
Nothing in the Linux kernel says that it requires a min SysVinit in the 
initrd  even the initrd's examples from kernel.org don't have it.   For 
Upstart to be complete it has to replace this part as well.   Then 
SysvInit scripts will no longer be needed for anything.  Only upstart 
system.  It does make editing initrd simple it being the same outside.  
Ok I need to setup a raid on startup expand the initrd  do the same 
settings as you would with SysVinit pack the initrd and you are done.  
Normally done when restoring from backup on different hardware and don't 
want to reinstall..  If upstart is truly replacing SysVinit it should 
replace the min form in initrd's as well.  Just so it stays the same.

So Upstart has not replaced all the mess left over from the SysvInit 
time yet.

Inside the Initrd is the deepest point in Linux the SysvInit style 
construction has got.

Basicly it started off at SysvInit and went deeper.  Then basics of 
SysvInit endup in stuff like hal and the like.   Coders recycle ideas.  
And if starting idea is bad if its not fixed early it ends up spread 
everywhere.

Peter Dolding