from Hacker News

Reinventing Home Directories – systemd-homed [pdf]

by kragniz on 9/20/19, 10:47 AM with 12 comments

• by LinuxBender on 9/20/19, 5:14 PM

  I am not comfortable with the feature creep that systemd is slowing bringing in. binfmt mount executing files based on a lookup, similar to windows assoc, intercepting gethostbyname(), this really feels like an attempt to shim old vulnerable windows concepts into linux. I am concerned that linux will not be recognizable in the near future and will be subject to bloat and obfuscation. What would it take to make this stop?

  There are some things brought in that I like, but I don't believe systemd is required for them. Cgroups, LXC, to name a couple. Those make functional and operational sense to me.

• by danmg on 9/20/19, 1:50 PM

  I'm going to pass on this one. How is /etc not 'extensible'? You just add a new file to the directory. Reinventing the windows registry, but this time in JSON, is a lateral move that is only going to complicate maintaining systems.

  For user level configuration, you already have too many competing mechanisms for doing this already, and this one requires systemd level integration making it much more coupled to the operating system and system in general.

• by hiciu on 9/20/19, 11:31 AM

  talk / context: https://streaming.media.ccc.de/asg2019/relive/164

  edit: I think the most important part of the talk was an answer to question about why .ssh/authorized_keys won't be possible with this scheme (42:00):

  > this is about protecting your data from the system as much as possible, so that when you are not logged in it's really cut off

  This looks like, among other things, an ecryptfs replacement.

• by airencracken on 9/20/19, 4:15 PM

  This seems like a massive boondoggle and a potential security issue.

  No thanks.

• by philpem on 9/20/19, 11:27 AM

  What is with this fetish of putting everything in systemd?

  Soon it'll have an X-server and a word processor built in...

• by mongol on 9/20/19, 8:36 PM

  Some of Lennart's inventions that he presents at conferences seem to lose momentum soon after. A while ago he talked about casync and mkosi. Both of these seemed more interesting to me than this but progress seem slow or stalled. We will see what becomes of this, but I don't think it is a slam dunk and what everyone has always needed but never realized. I think by ruling out remote /server use cases he will limit the interest in it.