On Sun, 2005-05-22 at 16:28 +0200, Julien BLACHE wrote: > Thomas Hood <[EMAIL PROTECTED]> wrote: > > > Version: 1.0.15-10 > > Duh, your mirror is fast.
I installed from incoming to see if the bug had been fixed in the latest release. > > MAKEDEV should be run even if udev is installed and active. In that > > case MAKEDEV will create a device node in /dev/.static/dev if > > I thought this wasn't the case anymore ? The current sarge MAKEDEV does a cd into /dev/.static/dev if the latter exists and so does /dev/.udevdb/. > > possible. That is good, because it means that the node will exist and > > be usable if udev is later disinstalled. Thus the intent of this code > > is wrong and even a policy violation (10.6). > > Uh ? How so ? Use udev or don't use it, in any case if something > breaks it's up to you to fix it. If MAKEDEV is called even when udev is installed then the breakage is avoided. > (moreover the device nodes created in > the postinst are standard and created by the installer -- the makedev > calls are just here for safety and older installs where the devices > might not exist, especially parport.) OK, I didn't know that. So this bug is _really_ severity: minor. > > However, I think that we have been lucky. udev no longer creates > > /dev/.udev.tdb when it is active; instead it creates the directory > > /dev/.udevdb . Therefore the "! -e /dev/.udev.tdb" test should succeed > > even on systems whereon sarge udev has been installed. I haven't > > checked a system on which sarge was freshly installed, but at least I > > can say that on my mostly-sarge-system-with-udev there is no > > /dev/.udev.tdb file. > > Yeah, OK, good, I'll just remove the check in -11 or -1 and wait until > another udev user reports a bug telling me everything is broken. No one will report it because the code doesn't misbehave. I just happened to notice the problem will looking at the code for other reasons. > Eh, udev was supposed to solve all the problems we had with devfs, > which was supposed to solve all the problems we had with static > /dev. Wait... what were those problems already ? :) -- Thomas Hood <[EMAIL PROTECTED]> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
