systemd-homed in Arch Linux

xerxes_ · 2019-12-07 14:47:11

My basic question is about future of systemd-homed in Arch Linux. When it will be available and working in future releases of systemd, will users be forced to use it, or will they have a choice?

WorMzy · 2019-12-08 15:10:11

Like most things systemd ships with (e.g. timesyncd, networkd, systemd-boot, systemd-firstboot, etc.), it's optional.

See https://media.ccc.de/v/ASG2019-164-rein … ies#t=2591 (notably ~43:00 for a direct answer to your question)

graysky · 2019-12-09 15:07:26

WorMzy wrote:

Like most things systemd ships with (e.g. timesyncd, networkd, systemd-boot, systemd-firstboot, etc.), it's optional.
See https://media.ccc.de/v/ASG2019-164-rein … ies#t=2591 (notably ~43:00 for a direct answer to your question)

Thanks for the link, that dude might be brilliant but damn, he says "umm" way too much!

dmerej · 2019-12-12 19:51:09

I for one am quite eager to try it

sekret · 2019-12-15 14:42:24

graysky wrote:

Thanks for the link, that dude might be brilliant but damn, he says "umm" way too much!

In his defense: He's from Germany and we germans say "umm" (in german "ähm") all the time. In my opinion he's doing very well! Remember, he's doing a talk in a non-native language, which is hard enough.

graysky · 2019-12-15 15:05:26

@sekret - Fair enough... if he still does it in his native language, this could be a drinking game!

Last edited by graysky (2019-12-18 12:18:30)

sekret · 2019-12-18 12:14:53

Doing so would kill you if you listened to me talking!

eschwartz · 2019-12-18 17:00:10

xerxes_ wrote:

My basic question is about future of systemd-homed in Arch Linux. When it will be available and working in future releases of systemd, will users be forced to use it, or will they have a choice?

I can't imagine why anyone would think Arch Linux would refuse to support any installation that does home directories the way all Unix systems do and have done it for decades. Even systemd upstream doesn't have that much... nerve.

Equally, I don't see any reason to believe Arch Linux would deliberately disable some inert binaries you don't need to use. It's not Arch Linux's job to tell people that homed is pointless, people can figure that out on their own.

ewaller · 2019-12-18 17:10:09

eeschwartz,

I don't often disagree with you; and I am not a systemd hater. But consider whether following statement might have been prescient a couple years ago :

"I can't imagine why anyone would think Arch Linux would refuse to support any installation that does process 1 the way all Unix systems do and have done it for decades."

Edit: Fixed the init process number after reading eschwartz's response O_o

Last edited by ewaller (2019-12-18 18:09:25)

eschwartz · 2019-12-18 17:56:52

There's a big difference between not supporting alternative init stacks for pid1, and dropping support for directories being "directories" instead of autogenerated partition mount points with goop on top, which even so will still show up as "directories".

Like, I don't understand what could potentially be enforced here. Is logind going to suddenly start including the following pseudocode:

if user_account.is_not_daemon_user() and user_account.home_directory.is_not_homed():
    error("if you're a login user, you are required to store your home directory in systemd-homed, either use homed or reconfigure your account as a system daemon")

Home directories on Unix are ridiculously lenient, due to the fact that they're just... random directories associated with a user account lookup.

Last edited by eschwartz (2019-12-18 17:59:30)

ewaller · 2019-12-18 18:17:09

I dunno. There is a lot of PFM technology in systemd. I still don't fully grasp everything going on with logind, seats, sessions, Xorg session and pulseaudio interaction. It works for me, so I don't delve into it.
Frankly, I don't dismiss the possibility that your assertion might become true.

Edit: Funny : "Home directories on Unix are ridiculously lenient" I read that as lennart

Last edited by ewaller (2019-12-18 20:28:42)

antiarch · 2020-02-05 00:10:14

Should this not be in Arch/Systemd? Is systemd linux? No?

fukawi2 · 2020-02-05 01:43:13

antiarch wrote:

Should this not be in Arch/Systemd? Is systemd linux? No?

Take the hint from your closed thread; don't be a troll. Have a 3 day ban to consider your position.

EDIT: actually, have a permanent ban for registering with a disposable email address, and 2/3 posts being trolling.

ngoonee · 2020-02-05 03:10:02

fukawi2 wrote:

antiarch wrote:
Should this not be in Arch/Systemd? Is systemd linux? No?
Take the hint from your closed thread; don't be a troll. Have a 3 day ban to consider your position.
EDIT: actually, have a permanent ban for registering with a disposable email address, and 2/3 posts being trolling.

The name was a bit of a giveaway?

headkase · 2020-02-05 06:01:48

As a modern initialization system systemd is fine. And it will continue to improve, many people make the mistake of being critical on initial impressions. Being relatively new it doesn't have to worry so much about maintaining a "legacy compatibility" and also is written from the base with lessons learned from previous systems. It is good enough that enough distributions are adopting it. And what is best about that is that it represents a "convergence" of things. Enough distributions have adopted it that it is becoming "the" standard. Cross-distribution standards like this facilitate software interoperability and as such are very desirable so I see systemd, in the wider ecosystem, just being a good thing. systemd is literally standardizing what used to be fragmented.

ewaller · 2020-02-05 15:02:22

headkase wrote:

Cross-distribution standards like this facilitate software interoperability and as such are very desirable so I see systemd, in the wider ecosystem, just being a good thing. systemd is literally standardizing what used to be fragmented.

Except that (IIRC) it is specifically targeted at Linux. Again, I am not a systemd hater, I like it and use it (otherwise I would be back on Gentoo). But, it does drive a wedge between the Linux and BSD based distributions. I'd prefer to work with them.

Head_on_a_Stick · 2020-02-05 16:54:18

It's a nice feature but 21,000 lines of extra code? That seems like a lot.

headkase · 2020-02-06 00:02:20

ewaller wrote:

headkase wrote:
Cross-distribution standards like this facilitate software interoperability and as such are very desirable so I see systemd, in the wider ecosystem, just being a good thing. systemd is literally standardizing what used to be fragmented.
Except that (IIRC) it is specifically targeted at Linux. Again, I am not a systemd hater, I like it and use it (otherwise I would be back on Gentoo). But, it does drive a wedge between the Linux and BSD based distributions. I'd prefer to work with them.

That is something I didn't consider, I have no BSD experience at all. Is there some technical fault keeping systemd out of BSD, or is it rather a political/opinion problem?

fukawi2 · 2020-02-06 03:44:47

headkase wrote:

Is there some technical fault keeping systemd out of BSD, or is it rather a political/opinion problem?

systemd makes heavy use of cgroups, which BSD does not have. IIRC, there are technical and/or political reasons why it never will either.

eschwartz · 2020-02-06 15:40:15

systemd does not work on alpine or void, both of which are linux distros. Currently prevailing opinion in those communities is that systemd is either evil, or bloat, or simply doesn't fit their goals for simplicity... but in fact, at least voidlinux at one point *wanted* to support systemd, and tried sending in patches to make systemd work on voidlinux. Those patches were rejected, they gave up, and eventually they decided it wasn't even worth the fuss, and "it's all for the best".

The problem? systemd explicitly relies on glibc, you cannot use alternative standard C libraries. Alpine is based on the musl libc, and voidlinux supports both musl libc and glibc so they cannot have core bootup sequence components which only compile on half of their supported platforms.

BTW: The *BSDs don't use glibc either.

The patches in question were rejected for political reasons.

sabroad · 2020-02-06 17:07:47

eschwartz wrote:

The patches in question were rejected for political reasons.

It was "political" only in the sense that the projects have different goals.

Lennart Poettering wrote:

> The problem is that musl is aiming at making a lean and efficient posix libc, not a glibc clone.
[...]
We will not take part in this game. Yes, we tend to gravitate more towards POSIX interfaces, if both a GNU and a POSIX version exists and both are equivalent in their functionality. But also: we will make use of GNU extensions and of Linux extensions if they are useful. And since these ultimately are part of the Linux API, that is the right thing to do.

Last edited by sabroad (2020-02-06 17:09:11)

eschwartz · 2020-02-06 17:17:31

sabroad wrote:

eschwartz wrote:
The patches in question were rejected for political reasons.
It was "political" only in the sense that the projects have different goals.

That's the literal definition of political, so I don't understand the point you are trying to make.

Last edited by eschwartz (2020-02-07 00:06:18)

sabroad · 2020-02-06 17:41:15

eschwartz wrote:

That's the literal definition of political.

Political means relating to the way power is achieved and used.

eschwartz wrote:

The patches in question were rejected for political reasons.

Rejecting these patches was not a power play.

eschwartz · 2020-02-06 17:51:57

sabroad wrote:

eschwartz wrote:
That's the literal definition of political.
Political means relating to the way power is achieved and used.

That's correct, the power to make decisions about goals lies with the project maintainers.

Who exercised that power.

"power".

Power was exercised, for political reasons like "this is not our goal" rather than technical reasons like "we cannot get it to work" or "it would break existing supported systems".

Returning to the original statement which you seem to be bizarrely objecting to:

eschwartz wrote:

headkase wrote:
That is something I didn't consider, I have no BSD experience at all. Is there some technical fault keeping systemd out of BSD, or is it rather a political/opinion problem?
The patches in question were rejected for political reasons.

I reiterate, this was a political reason.

sabroad wrote:

eschwartz wrote:
The patches in question were rejected for political reasons.
Rejecting these patches was not a power play.

You have artificially introduced the phrase "power play" into this conversation for reasons I cannot fathom.

If you are trying to play lawyer word games, please stop.

Last edited by eschwartz (2020-02-06 17:52:38)

fukawi2 · 2020-02-06 23:58:41

eschwartz wrote:

That's the literal definition of political. Here, have an internet cookie as a reward: https://i.imgur.com/9C2fv49.jpg

Please consider your responses to other users before you hit "Submit". This is far from the first time your posts have come across as confrontational, intentionally or not. I don't expect you to agree with everyone, but I do expect discussions to be civil. You contribute a lot of time and effort to Arch, and hold a position of respect in our community, including from myself, but the perception of these kinds of posts is discouraging to others, and not in the spirit of the community we're trying to foster in the forums. Please consider good faith on behalf of those you're interacting with before you start tearing apart the minutiae of their posts.

Arch Linux

#1 2019-12-07 14:47:11

systemd-homed in Arch Linux

#2 2019-12-08 15:10:11

Re: systemd-homed in Arch Linux

#3 2019-12-09 15:07:26

Re: systemd-homed in Arch Linux

#4 2019-12-12 19:51:09

Re: systemd-homed in Arch Linux

#5 2019-12-15 14:42:24

Re: systemd-homed in Arch Linux

#6 2019-12-15 15:05:26

Re: systemd-homed in Arch Linux

#7 2019-12-18 12:14:53

Re: systemd-homed in Arch Linux

#8 2019-12-18 17:00:10

Re: systemd-homed in Arch Linux

#9 2019-12-18 17:10:09

Re: systemd-homed in Arch Linux

#10 2019-12-18 17:56:52

Re: systemd-homed in Arch Linux

#11 2019-12-18 18:17:09

Re: systemd-homed in Arch Linux

#12 2020-02-05 00:10:14

Re: systemd-homed in Arch Linux

#13 2020-02-05 01:43:13

Re: systemd-homed in Arch Linux

#14 2020-02-05 03:10:02

Re: systemd-homed in Arch Linux

#15 2020-02-05 06:01:48

Re: systemd-homed in Arch Linux

#16 2020-02-05 15:02:22

Re: systemd-homed in Arch Linux

#17 2020-02-05 16:54:18

Re: systemd-homed in Arch Linux

#18 2020-02-06 00:02:20

Re: systemd-homed in Arch Linux

#19 2020-02-06 03:44:47

Re: systemd-homed in Arch Linux

#20 2020-02-06 15:40:15

Re: systemd-homed in Arch Linux

#21 2020-02-06 17:07:47

Re: systemd-homed in Arch Linux

#22 2020-02-06 17:17:31

Re: systemd-homed in Arch Linux

#23 2020-02-06 17:41:15

Re: systemd-homed in Arch Linux

#24 2020-02-06 17:51:57

Re: systemd-homed in Arch Linux

#25 2020-02-06 23:58:41

Re: systemd-homed in Arch Linux

Board footer