On Thu, Apr 02, 2015 at 03:09:49PM +0200, Andreas Henriksson wrote: > Hello Josh Triplett (and #753779)! > > I've just spent a few minutes revisiting the idea of shipping mountpoint > in util-linux. I'll do my best to make that happen as soon as Stretch > development cycle opens up.
Thanks! > I'm guessing your interest in this is that you might want to purge the > "initscripts" package? Yes, exactly. In particular, mountpoint seems to be the main thing that makes initscripts pseudo-essential. > I've looked a bit at right now and that seems a > bit more problematic. The reason packages depends on initscripts is not > only mountpoint, but also because of LSB init script header > dependencies. > See for example https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=581420 > which talks about this for the hwclock.sh init.d script shipped in > util-linux. There are also several other packages shipping an init > script with LSB header depending on init scripts from "initscripts". > (systemd itself might be able to drop its initscripts dependency, but > other packages in the base system will still pull in initscripts for > you. c.f. procps, etc.) > > At first I had the idea that, ok maybe all packages can just change > to depend on 'initscripts | systemd-sysv'. Unfortunately that can > break the case where you have systemd-sysv installed but > boot with "init=/lib/sysvinit/init" on the command line (and not > have initscripts installed). I think that could be fixed simply by moving the initscripts dependency from sysvinit-core to sysvinit. If you have sysvinit installed, and you want it to be a functional init, you need initscripts (and potentially other dependencies that currently only appear on sysvinit-core). And if you *don't* have sysvinit installed, just systemd (and systemd-sysv), then systemd just needs to provide all the necessary LSB-required init script names itself, which in the future it can do without depending on initscripts. At that point, I think a package that provides an init script could then stop depending on initscripts (unless it needs a versioned dependency), and only depend on a functional init (which is essential, so it doesn't need Depends). (For the duration of stretch, such packages would also potentially need a versioned Breaks or similar on versions of sysvinit that didn't have the initscripts dependency.) > While I'm interested in having a "legacy-free system" (no sysvinit parts > lingering) I think accomplishing that might be hard to implement in a > way thats acceptable for Debian or atleast too much trouble for what > it's worth. If you're interested in coming up with a solution for this > that would be very welcome! I think the above proposal of making sysvinit depend on initscripts should solve that problem. I filed a bug on sysvinit to that effect. > PS. I'll be closing this bug as soon as /bin/mountpoint is shipped > by util-linux. Feel free to file a separate bug about initscripts > dependency if you're interested, but please include a way to actually > solve getting rid of the dependency if you do file it. Absolutely; this bug is *only* about moving mountpoint to util-linux. - Josh Triplett -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
