On Jul 13, Josselin Mouette <[EMAIL PROTECTED]> wrote: > Then I guess you are deliberately breaking all up-to-date unstable > boxes? In this case udev installs, but doesn't start at boot time > anymore, while with 0.060-1 it starts. I decided that this would have been the less painful choice, having udev disabling itself if there is the concrete risk that the system will start with an half-populated /dev. If instead it worked fine for them users can edit the init script for a few days (it's still not obvious why it breaks, and it has even be suggested that this is a generic bug not related to kernel versions).
> > > Oh, and please stop this madness. There aren't any 2.6.12 kernels in > > > unstable, there may not be for a while, and not everyone wants to lose > > Actually members of the kernel team hinted that they could be available > > in a couple of days. > > After fixing #317720 the situation of udev is not really different from > > libaspell, unstable is not guarantee to be installable the 100% of the > > time. (Actually I was wrong, on a fresh install the package will install but automatically disable itself.) > However unstable packages should always be release-quality. I wouldn't "Should" is the keyword here. It should have been, but it ended up buggy. This was not supposed to happen, when I uploaded udev 0.060 I could not know that it would break with 2.6.11 kernels. > Then why are you refusing to support earlier kernels like it has been > suggested? Because a general dual-personality package is not practically possible, as I explained multiple times. The newer udev versions use different semantics for the rules files and provide different features to other components of the system, so switching back and forth between versions is not practical. -- ciao, Marco
signature.asc
Description: Digital signature
