On Sun, Apr 29, 2007 at 08:54:56AM -0700, Andreas Kabel wrote:
>> Well, one cannot simply move the entire world into /.
> Of course not. I was specifically suggesting rpc.idmapd and rpc.gssd,
> that's hardly the world.
But rpc.gssd depends on a full host of Kerberos libraries, and possibly SPKM3
after a while. rpc.idmapd depends on SASL, and in turn GnuTLS, libldap and
tons of other stuff. You could of course argue that it's "only" 5MB of
libraries, but I'm not sure if you could justify moving that much stuff into
/lib for bootstrapping purposes.
> Well, one cannot simply declare every bug to be a feature. What exactly
> is the rationale behind not having /usr on nfsv4 and having scripts
> fail with a catastrophic error when you try to?
The problem that rpc.{gssd,idmapd} is in /usr? :-)
I wonder if the best solution here would be making the initramfs capable of
running some sort of minimal bootstrapping environment for NFSv4 -- I've
heard rumours that the klibc people are looking into NFSv4 on /, and in that
case, you could probably solve both / and /usr in one go (at least
theoretically). After all, bootstrapping is what we have initramfs for...
/* Steinar */
--
Homepage: http://www.sesse.net/
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
