Control: severity -1 wishlist Control: tags -1 wontfix Control: close -1 On Fri, 22 Jul 2022 22:29:22 +0200 Adam Borowski <kilob...@angband.pl> wrote: > Control: reassign -1 udev > Control: retitle -1 udev: Please drop systemd from Depends > > > Control: reassign -1 sysvinit-core > > Control: retitle -1 sysvinit-core: please depend on systemd- standalone-sysusers, systemd-standalone-tmpfiles > > > > On Fri, 2022-07-22 at 21:19 +0200, Adam Borowski wrote: > > > And if you want a real package first, -standalone is a clear winner > > > here. > > > > That sounds too fragile for packages involved in debootstrap. It seems > > more reasonable to have sysvinit-core pull it in for those who care as > > was suggested earlier. > > Sorry, there is no way to do so in sysvinit. Sysvinit itself doesn't depend > on udev thus it _really_ doesn't want a 450KB extra in tiny containers -- > and there are many other inits: runit-init, all those dumb-init finit-sysv > pid1 tiny, see my .sig; then it's fully valid and supported to run without > any init at all. > > The fix must be done in the package that regressed -- ie, udev. > > And I really don't appreciate adding 450KB to small containers; reducing > their size is something many of us is fighting for for many years.
The udev dependency is generated correctly, there is no regression. What you said about "switching init system" is incorrect, merely installing the systemd binary package does no such thing. It does not make alternatives harder to use in any way, in fact it makes precisely zero difference. If you want to influence how the alternative is chosen, you can add a dependency to sysvinit-core as suggested (or a breaks), or use debootstrap's --include option, or any number of other solutions. -- Kind regards, Luca Boccassi
signature.asc
Description: This is a digitally signed message part
