On Wed, Sep 20, 2006 at 11:52:29AM +0200, maximilian attems wrote: > On Wed, Sep 20, 2006 at 11:06:54AM +0200, Sven Luther wrote: > > On Wed, Sep 20, 2006 at 11:00:21AM +0200, maximilian attems wrote: > <snipp> > > > nono, not debconf please. > > > > Why not debconf ? > > useless user intervention, useless work for translators. > not portable beyond Debian unless carefully done.
Well, for 1) it is clear that it would have a sane default, andprompt only at medium or low priority, which takes care of that, for 2), well, i have no strong opinion on this, but i guess there is not so much of a problem, not sure though. As for 3), it is evident that it should be well and carefully done, so your critic anihilates itself :) > <snipp> > > > the idea is to backup on an update call any initramfs > > > that is older than a certain time, lets say 6h or maybe 24h > > > thus not created while upgrading mdadm, udev, usplash, cryptsetup or > > > so together. > > > > This will not guarantee you that you backup the right one, and doesn't work > > in > > coordination with the bootloaders. > > > > > will be in 0.81 update-initramfs. > > > > Ok, but i think it is a partial solution and misses the complete problem. > > no but thanks for your pointer. > so keeping a back up wile running is a good idea, dpkg.bak. > if there exists no young .bak it will be saved there. > so powercuts are catches for block based bootloaders too and > more modern bootloaders find a .bak that has a known old state, > presumed good. Yeah, maybe i will find time in the next days to think aboutthe whole solution, and make a full proposal (perhaps on the wiki page ?), so we can discuss it ? Not sure if it will be an etch thingy or not though. Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
