On Wed 16 Sep 2020 at 16:15:12 (-0700), Patrick Bartek wrote: > On Wed, 16 Sep 2020 13:52:15 -0400 > Greg Wooledge <wool...@eeg.ccf.org> wrote: > > On Wed, Sep 16, 2020 at 10:32:14AM -0700, Patrick Bartek wrote: > > > To make a long story short, after two or so weeks of research and > > > numerous failed trials, I came to the conclusion that systemd has > > > become too entrenched in the dependency tree of Buster to successfully > > > convert to systvinit. > > > > If you specify "... on a desktop system", then maybe you're correct. > > > > For most servers, it shouldn't be an issue. > > The subject _was_ about desktops, MATE specifically, not servers. > > However, my trials with Buster was from a year ago. And I haven't > tried a sysvinit install with it since. Perhaps some systemd > dependencies have been eliminated. Be great if they all were! Init > systems should never ever be dependencies.
I know little to nothing about DEs. However, I see that there are people who run MATE without running a systemd init system. This (dated) link makes a distinction between installation dependencies and runtime dependencies, so I presume that you might be able to put up with the presence of unused systemd packages in the installation. https://unix.stackexchange.com/questions/279603/using-mate-desktop-without-systemd Later: > […] Had no problems converting to > sysvinit with a terminal only system. First thing I did. I always > start my installs that way and build from there. Lighter, faster, more > efficient system without all the crud that comes with a general DE > install. I would certainly recommend that the OP did that, rather than converting as an afterthought. Cheers, David.
