On Sun, Aug 17, 2008 at 01:36:09AM +0200, Marco d'Itri wrote: > On Aug 17, Andrew Pollock <[EMAIL PROTECTED]> wrote: > > > So from my reading of the bug report, there appeared to be a chicken-and-egg > > bootstrapping problem, which would be resolved by udev creating /dev/random > > (and/or /dev/urandom). > Which I really do not want to do, because it opens the way to every > "special" (read: "retarded") application to request that its own device > nodes are manually created to work around some misfeature. > Can you make the case for this situation to be more special than others?
I can't think of that many other devices than /dev/[u]random, which would fall into this category. In this case, you've got LDAP (over TLS) providing NSS information, which udev presumably needs in order to start. This is one of those very low-level cases. I'm failing to think of any other, non-system type use-cases where udev would need to create the device. Can you? regards Andrew
signature.asc
Description: Digital signature
