Re: A newbie's view

[Kenneth Skaar <Kenneth.Skaar@runit.sintef.no>]

On Thu, 28 Jan 1999, Seth M. Landsman wrote:

> 
> > > 	I think, besides have lilo backup a working kernel, the make
> > > process should do some sort of integrity check on the kernel to make sure
> > > it would boot (i.e., check that the elf loader is in there).  It isn't
> > > very useful to tell the user that this failed if you don't given them some
> > > clue how or why it failed.
> > 
> > This sounds like a good idea.
> > 
> > Definitely should be added to the list.
> > 
> > Although a lot of this sounds like an aftermarket extra - Some sort of build
> > script/interface that is very easy to deal with. and very tolerant. and
> > cautious.
> 
> 	This definitely shouldn't be part of the kernel proper (i.e., the
> kernel that is compiled should be identical after these changes go in).
> But I do think things like this should be part of the kernel makefile ...
> 	I think for any complex system, it becomes increasingly necessary
> to have some sort of good configuration and testing facility.  I think
> that the kernel is lacking in both.
>
> -Seth
> --
All of this is currently beeing worked on by the Linux Kernel Manager
project(http://www.caribdata.co.uk/psi-domain/lkm.html). They haven't been
around for long, but they are moving forward, and it seems like their
doing The Right Way(tm). I dont know how they plan to keep up with all the
new configuration-options that pops up now and then, but I guess they'll
read and parse the source config-files, and then offer wizard-style help
for the "known" parts, and just plain kernel-build-help for the new stuff. 

Anyway, IMHO keeping a "user" app like this out of the kernel source
tree is a _good_ idea. 

just my 2 kr..

Kenneth 

-
Linux-future: thinking about the future of the Linux kernel
http://humbolt.nl.linux.org/lists/