On Thursday 16 June 2011 15:44:58 Kay Sievers wrote:
On Thu, 2011-06-16 at 15:35 +0200, Anders Johansson wrote:
On Thursday 16 June 2011 15:21:29 Kay Sievers wrote:
What fails? Can you give some relevant examples?
Everything that uses the udev database which was populated with /usr around but needed tools from there to do so.
Ah yes. udev is also one of your projects. Also pushed into the initrd for absolutely no reason whatever, and the cause of so many problems because of it that it's not even funny
Thanks for the absolutely-no-reason-whatever expert opinion.
You're welcome. I can show you the pain it has caused me and my colleagues. Because of udev in the initrd, we now have discovery of every single device so early, we have problems with the interaction between mpio and lvm/md, we have problems with configuration files because every time anything changes that affects those things the initrd has to be rebuilt, we have problems debugging because it's virtually impossible to debug discovery failures in the initrd. The initrd is supposed to be an extremely small thing that gets the root file system mounted, and everything else is supposed to happen afterwards, once the real root has been mounted. It is not supposed to contain a small operating system And yes, udev in the initrd is the cause of quite a number of totally unnecessary bugs. Especially when it was pushed into the initrd in SLES9 while it was still deemed unfit for the main system, but even today Anders -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org