SquashFS?

[Jeremy Katz]

On Fri, 2005-10-21 at 21:49 +0100, Luciano Miguel Ferreira Rocha wrote:
> On Fri, Oct 21, 2005 at 04:21:39PM -0400, Jeremy Katz wrote:
> > On Fri, 2005-10-21 at 19:23 +0100, Luciano Miguel Ferreira Rocha wrote:
> > > On Fri, Oct 21, 2005 at 08:03:16PM +0200, Darko Ilic wrote:
> > > > Well, having unionfs included in the kernel would be *great* for live CDs. 
> > > > 
> > > > If we could push both unionfs and SquashFS to go upstream, that would improve 
> > > > the quality of live CDs dramatically. 
> > > 
> > > To have unionfs and squashfs upstream would be nice, yes. But they can
> > > be used regardless of being upstream or not.
> > 
> > Not for an official Fedora Live CD.  One aspect is that it *must* be
> > built using components distributed as part of Fedora and we're
> > (generally speaking) against patches which aren't upstream because they
> > significantly raise the maintenance burden and then also get people
> > complaining because the kernel isn't "stock"
> 
> I was thinking of adding the source of the modules to the "official
> fedora livecd maker". The _binary_ rpm would have the sources.
> 
> Thus, no need for kernel patches or any unionfs/squashfs package at all.
>
> The fedora livecd build process will compile the modules on livecd
> creation, using the kernel-dev package for the kernel to be used by the
> livecd. As the livecd system won't be updated anyway, there isn't even a
> need of keeping any unionfs/squashfs package at the same level of the
> kernel package.

And if the compiler on the system is different?  This is a really
dangerous path to go down.  Also, unionfs and squashfs have to be at the
same level as the kernel package because that's where the kernel for the
live cd comes from.

Jeremy