On Apr 04, martin f krafft <madduck@debian.org> wrote: > BTW, that udev maintainer, with all due respect, could really talk > to me and/or the bug report about these things. I cannot discuss bugs I do not know about, somebody talked about using udevtrigger for this on some channel and I explained how it would be wrong. I did not understand that it had actually been proposed to you as a bug fix. I am available on IRC about 15 hours every day if you want to talk about issues related to my packages, which you should do since you do not appear to be familiar with it. > That udev maintainer > could also explain how a script in /sbin with a manpage and all does > not constitute a public interface. Actually I explained that it is not correct to synthesize all events multiple times at boot time, because it is wasteful and may confuse other programs. But still, even if a binary is documented it does not mean that it should be used by other packages disregarding the consequences on the rest of the system... *If* you really have to synthesize some events at boot time (which is not obvious, since AFAIK this has never been correctly supported until it was fixed in recent kernels as the udev upstream maintainer explained to you in january) then you must do it only for the md* events by manually poking the uevent files: for device in /sys/block/md[0-9]*/; do echo add > $device/uevent; done -- ciao, Marco
Attachment:
signature.asc
Description: Digital signature