Re: initrd boot

[jfm2@club-internet.fr]

> Date: Thu, 22 Jan 1998 16:49:19 -0800 (PST)
> From: George Bonser <grep@oriole.sbay.org>
> Sender: grep@calvin.shorelink.com
> cc: rickya@siservices.net, k.wetzel@welfen-netz.com, starsend@interlog.com,
>         seul-dev-distrib@seul.org
> X-No-Archive: yes
> Content-Type: TEXT/PLAIN; charset=US-ASCII
> X-UIDL: 28b25416379520f515b0b970f53b8686
> 
> 
> initrd is something that CAN improve the boot process at some point in the
> future. I am simply trying to keep folks focused on the basics at this

I don't see why we should wait for using a stable, non beta feature
who allows to build far leaner kernels thus reducing the need for the
user recompiling the kernel.  A user with 8 megs _cannot_ live happy
with the kernels over 2 Megs in size shipped in some distributions.
And about the cost of having every SCSI driver compiled in the kernel
it is over 500K.  Of course this will be bigger in 2.2.


> point. 
> 

The central point is than SEUL is _not_ a distribution where we can
say: "No problem the user will recompile the kernel."  With 2.0 it is
_already_ possible to have distributions where recompiling the kernel
would be nearly useless.  But some distributions are shipping with fat
kernels because it makes their work easier and don't consider kernel
compilings be a problem for _their_ users.


> 1) We need to define SEUL and the Linux Core in an official document.  
> 2) We need to modify Debian's current system to provide for a SEUL
> install.

Yes.  I found than Debian is made by hackers and than at times they
are much too prone to ignore normal user problems.

> 3) We need to begin to identify
> 	A) What is difficult about the current install and how to we fix
> 	it.
> 

Most of us have been thinking about these problems for months.  In my
particular case I remember how autumn 95 I was unable to recommend
Linux to my brother in law, one of the decisive factors being the
kernel recompilings mandatory in 1.2.13.  It was then than I began to
investigate choke points in the use of Linux and how distributions
tried to remove them.  I think than the older and more active people
in this list have been doing the same thing for months or years.  We
are not _beginning_ to identify, we are beginning to _act_.

I kept a close watch on the 1.3 kernels and had high hopes in the
nearly complete modularization.  When 2.0 was released I found than
distribs didn't adapt really to it on the basis the user can recompile
and in fact kernel size is a very good indicator of how much a
distribution cares about end users.

-- 
			Jean Francois Martinez

==================== The Linux.  Use the Linux, Luke! =======================