[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: SEUL: Reuse vs. new code (was Re: Idea for SEUL startup)



On Sun, 24 Aug 1997, Erik Walthinsen wrote:

> >	1- Ditch sysvinit, write a new init.
> 
> While the goals listed below (in the original message) are reasonable, the 
> method isn't:

[compelling reasons not to ditch sysvinit removed]

You may find this argument unrelated to the sysvinit thing, but I don't. 
I'm also in favor of overhauling sysvinit to some extent, such as
scrapping some of the scripts for a more reliable binary "daemon manager". 
But my true argument is this: the boot process will have to be made a bit
more like win95.  The runlevels already symbolize things like "Normal",
"Safe Mode", "Command prompt", etc., to some extent, but that will need to
be improved upon.  LILO will have to go, it's simply too cryptic, ugly,
and rigid.  There will have to be some way to mask startup messages from
the user like win95 does with that logo screen.  Etc.  Etc. 

Masking boot messages from the user is important, but I can't tell how
to do it without adding some code to the kernel.  Maybe that's the best
approach.

Final note is on LILO.  I believe I've found a superior boot loader, both
for end-users and hackers alike.  Check out GRUB, the Hurd's boot loader,
from ftp://prep.ai.mit.edu/pub/gnu/gnu-0.2/grub-boot.image

GRUB doesn't need unreliable, easy-to-break map files, as it can
understand FAT, BSD, and Ext2 filesystems directly.  It can list anything
in context, including partition tables and directory contents, by hitting
<tab> at the right point.  It also starts out in a user-friendly menu with
an auto-boot timeout.

GRUB will need some work to be useful to SEUL, but a lot less than LILO, 
and it _is_ only at version 0.4.

-Sam Revitch