GNOME: Introducing stronger dependencies on systemd
Posted by small_kimono@reddit | linux | View on Reddit | 114 comments
LOL.
Q: So what should distros without systemd do?
A: First, consider using GNOME with systemd.
IamscaredForEars@reddit
I have no hate against systemd. But this is a bad thing imo.
Left_Security8678@reddit
How? Like 99,9% of Linux Users are on Systemd Distros.
WaitingForG2@reddit
Embrace
Extend
Extinguish <-(you are here)
nightblackdragon@reddit
What kind of EEE is that? systemd is open source and under free license. Do you even know what EEE means or it's just a term you are using when technology ju don't like replaces other technologies?
WaitingForG2@reddit
That makes Desktop Linux completely dependent on systemd by extinguishing concept of any other software in init/system manager space.
Controlled by corporation, so doesn't matter that it's open source or free licensed. You can't fork it even if you are long term maintainer with most commits, see XOrg fork. Corporation controls full direction of whats to come.
nightblackdragon@reddit
Yeah, you don't know what EEE means.
It's not controlled by corporation any more than things like Xorg, Mesa and more that you are using on your desktop.
WaitingForG2@reddit
https://en.wikipedia.org/wiki/Embrace,_extend,_and_extinguish
nightblackdragon@reddit
So you picked one sentence from EEE definitions and ignored the rest.
Jegahan@reddit
Man its so funny seing spread the same misinformation, even after they have been provided with evidence its BS.
You can absolutely fork systemd. Your comparison with the XOrg fork doesn't even make sense bexause you know that Xlibre exists. The guy was able to fork it, so what the hell are you talking about ?
You can also just go to the systemd repo and see that it has 4k forks.
WaitingForG2@reddit
How much traction it will get the moment systemd development will do something controversial?
For example, chromium is open sourced as much. I can't see hard fork that saved us from Manifest V3 though. Same will be here.
Oh, and about "misinformation". You are the one who spreads the word based on "from what you heard"
Jegahan@reddit
First off, this was your claim. Why are you now dodging to another bad example (chromium), when your first was already proving that you are wrong, given that xlibre, a fork of X.org, exist? You even aknowledge it, right after claiming you can't fork it, contradicting yourself in one sentence. Secondly, nobody being willing to do the work to maintain a fork of a project doesn't mean that it can't be done.
Oh this is adorable. 1. The claim being based on something I read, doesn't make it misinformation. You could have asked me for a source, but you didn't, probably because you don't care about whether it is true or not. Its not like you would change your opinion, even when presented with evidence. 2. The source is the X11 gitlab where 7 month ago, the other devs where already complaining about metux's contributions causing more problems than they solved. This was long before the xlibre dev through a tantrum and wrote the readme of his fork. From the gitlab:
"Honestly, I would strongly recommend just not merging anything @metux does from now on. I do not feel that their presence here has been a net positive -- I have seen zero actual bugs solved by any of their code changes. What I have seen is build breakage, ABI breakage, and ecosystem churn from moving code around and deleting code. Xorg could use some actual maintenance, but that means fixing actual bugs and solving real problems."
You can go through the gitlab and find other examples of it (like this one 3 months ago. But I doubt you'll even read the example I just gave.
Left_Security8678@reddit
Cope. Systemd is the better init and ecosystem and thus destroyed the rest. Next time make better Software if you want to compete.
WaitingForG2@reddit
Not coping, just stating the fact.
This action is literally attempt to flip 99.9% closer to 100% by just making harder life to other inits/service managers. Next they should add systemd dependencies into GTK, that surely will be a good long term decision.
Okay. How it will make Gnome not depending on systemd though? Should Gnome change their dependency on init that is "better software" at current moment of time?
Left_Security8678@reddit
... yeah? Like the point is to use Features the Software provides to make things easier.
IamscaredForEars@reddit
Yes most are. But not all. I and agains no hate against systemd. But this kills a lot of portability. There are enough reasons not to use systemd. Maybe it doesnt fit in with your usecase, Maybe i just dont like the name. Hard dependency on other programs like this is bad
-o0__0o-@reddit
I wonder what chimera linux is going to do.
Left_Security8678@reddit
Finally make KDE their flagship
Tiny_Prune_4424@reddit
Would be funnier if they engineered around the SystemD dependency as an up yours to em
Left_Security8678@reddit
Manually patching software will eventually break.
Tiny_Prune_4424@reddit
Which is why I say if. KDE would be a pretty good replacement for GNOME when this goes through.
Fluid-Pirate646@reddit
https://gts.q66.moe/@q66/statuses/01JXFP9ESPBTWNDTZ00WD5T81N
losermode@reddit
Can anyone explain why there seems to be a lot of hate for systemd (and to lesser but still real extent, GNOME) among some Linux users?
Genuinely seeking to understand!
Tiny_Prune_4424@reddit
It tries to be everything but ends up being kinda bad at everything it does
Boots too slow, has too much baggage most won't use, crashes way more (at least for me), is harder to finetune as it isn't a shell script, adds lockin, etc etc
small_kimono@reddit (OP)
You might see: https://archive.is/506aS
IMHO the reason it is/was so disliked was wrapped up in one big philosophical dispute (which the systemd authors were ultimately correct about), and many small, petty tactical arguments (many of which the systemd authors were dead wrong and smug about).
KontoOficjalneMR@reddit
I knew SystemD maintainer was an egomaniac but this one takes the cake
"Sure, users might be bricking their motherboards and trashing hundreds of dollars worth of hardware by mistake, but I can't be arsed to remount file system RW when we very rarely need to write to it. So your bricked motherboards are sacrifices I'm ready to make".
ouch.
Krunch007@reddit
Systemd is still a bunch of related projects and modules rather than, as many portray it, one big monolithic bloated mess. Like, most distros don't bundle even half of systemd modules.
Some systemd projects don't even depend on systemd itself. For example, you can use systemd-boot on a systemd-free distro. You lose some logging features but it's otherwise just fine.
Richard_Masterson@reddit
That's disingenuous. Windows is a bunch of related projects and modules too, but you cannot use any of them without rewriting or reimplementing the rest which is virtually impossible.
The same thing applies here. Realistically nobody can reimplement the rest of systemd's modules to make it work without pulling the entire thing.
IncreaseOld7112@reddit
Im using systemd with kernel bootloader and not their dns. I don’t think I’m using their Bluetooth. I can’t swap out the windows init system.
mwyvr@reddit
Some of the components are, sadly full of crap code or bad design or too much scope.
There are perfectly legitimate reasons to dislike systemd without reaching for philosophical reasons. In general many of the ideas and goals are correct, but the implementation and Linux lock in are not.
NaheemSays@reddit
The interesting thing is systems is developed more like a Unix system (single repository) than most alternative systems that critics promote.
Ok-Salary3550@reddit
Yeah, to an extent systemd is just taking the BSD “one big system” concept and applying it to Linux.
losermode@reddit
That first article is pretty long but I like reading about things like this so I'll come back to it.
Thanks!
RedditMuzzledNonSimp@reddit
https://suckless.org/sucks/systemd/
OCPetrus@reddit
I've worked professionally on integrating systemd stuff at multiple companies for different use cases from cloud to embedded. Systemd is extremely opinionated. It solves most use cases very well, especially anything desktop and server. But in niche cases such as you can find in some embedded solutions it can get very painful to create workarounds.
What's worse is that it didn't have to be like this. Systemd would probably have been just fine with a less opinionated approach to many things. The issues were brought up during development. But the devs basically told users they don't care.
IncreaseOld7112@reddit
Can you be specific? Are you talking about the declarative nature? It seems to me it just maps really really well to the problem domain.
Richard_Masterson@reddit
For the very same reason of this thread's topic.
Systemd by design takes more and more things of the GNU/Linux OS. Back in the day the opponents claimed that the logical conclusion would be that DEs and other packages would end up with systemd as a hard dependency. Systemd acolytes claimed that it was a conspiracy theory, that systemd was just an init system and everything else would continue working as always.
Fast forward to today and GNOME will now have systemd as a hard dependency. The acolytes, of course, now claim that it was always the point of systemd and that it's actually a good thing.
Other reasons people disliked systemd were:
ABotelho23@reddit
Because new thing is not old thing.
ninth_ant@reddit
It’s absolutely not just familiarity, though I’m sure it’s part of it.
I’ve used systemd for much longer than I ever used earlier paradigms and I still miss the simplicity of the old ways. Even after many years of use both professional and at home, systemd just doesn’t feel like most other parts of Linux or UNIX, it’s big and heavy and obtuse.
Yes, I understand the benefits of systemd. In many (many) aspects the old ways were worse. Any grumbling I might have isn’t nearly enough to change to some niche distro or break a fundamental part of the distros I use on my systems.
luciferin@reddit
FreeBSD is probably your best option. All it takes for me to shake myself out of my reverie is remembering the first time I got stuck in a daemon dependency cycle on Archlinux/rc.d Yes, it felt nice to visualize what my system was running on boot from right to left in a simple text string, but ultimately that simplicity was a mirage.
mwyvr@reddit
On FreeBSD GNOME is three years behind current GNOME precisely because of GNOME's systemd and Linux dependencies. This stands to get worse due to the recent announcement by the GNOME project member.
Good job GNOME team.
Cry_Wolff@reddit
No one gives a damn about BSD.
mwyvr@reddit
What are you, a twelve year old fan boy?
Come back at me when you grow up.
luciferin@reddit
Honestly, these expectations are just unrealistic. No one is giving Valve shit because Steam won't run on Windows Vista. They have exponentially more resources than the GNOME team, and we pay them for some of their products. Why do we expect more of GNOME?
mwyvr@reddit
That comparison isn't at all relevant.
mwyvr@reddit
I recall mail list messages where other non Linux operating system developers worked to support GNOME with the delighted appreciation of the GNOME team then, going back to the very early days of GNOME.
It isn't the fault of other OS developers and port/pkg maintainers for becoming more insular and unnecessarily so, it is the GNOME team.
They have made choices that could be avoided.
ABotelho23@reddit
I think this is subjective.
The functionality systemd provides would be incredibly complex to implement in old init systems.
sparky8251@reddit
As someone that regularly messes with service management of custom programs at home and at work, init scripts are NOT simple and i hate the idea some people have that they are. They are miserable by comparison to 5 lines in a .service file, 2 of which are category markers, 1 of which is a description of the service in plain text...
Existing-Tough-6517@reddit
People hate change inherently so when you rip up the old people are quick to find fault and gnome especially gave people plenty of material to find fault with
I have neither time nor energy to write down the 85 different times they completely fucked up but I'll name a few.
Spent first 8 years of gnome 3 leaking memory until falling over because JavaScript runtime doesn't deterministically free memory while not acknowledging problems. Perment fix run gc like mad constantly marvel that performance isn't that bad.
Create no extension system just let extensions monkey patch ja that runs desktop extensions can break any given update and can crash entire desktop causing loss of all unsaved work. Did not learn from Firefox at all.
Hired scammer who stole from people charging them for fake energy healing to run gnome foundation
Took 18 years to implement previews in file picker
Leaders opined users shouldn't get theming because it would hurt "brand identity" of gnome if mom couldn't identify OS of son hiding in basement.
Systemd was just made by the punchable faces that brought you pulse audio
It did have that issue where any user called 0day could trivially promote themselves to root and the time Linus had to yell at them for breaking debugging but mostly it was just inherently complicated and different at the same time it was not trusted yet.
viva1831@reddit
I don't hate it - I'm just happier without it and don't like it enforced on me :)
It's been so long since I used it, about 10 years, but when I did I remember finding it janky (in my opinion), trying to do too much, and doing things in an annoying way. It felt like an annoyance or obstacle rather than a help. Back then we were also using a systemd spin-off in work (iirc named fleet?) which was horribly specified and just a nightmare to use. Unit files had so many limitations I sometimes ended up just having them call a shell script to do the work in any case. Command lines that could have been laid out neatly in a script became this nasty mess all crammed into one line
On a basic level I like my services to be launched by shell scripts. It's flexible, it's elegant. And every linux admin already knows the shell and shell scripts. Whereas for systemd, I have to learn a whole new format - meh
I don't like how it was essentially forced on many users, the attitude some folk had about it, or how it's now an operating in its own right, which is difficult to pick and choose components from
I do recognise this is entirely personal preference. I'm skeptical if "objectively better" is a real thing. I just like my way best. I'll continue to just not use it or any distro that enforces it. No point in hating - I'm happy to just quietly continue without it :)
sparky8251@reddit
You do uh... You do know you can have ExecStart run a script? So there are no messy one liners...?
viva1831@reddit
Wow that's one hell of an assumption. I won't discuss this with you if you veer into making it personal - there's no need to turn this toxic
sparky8251@reddit
So, explain why you couldnt just give it the path to the script, when thats been supported from the start and had to make ugly one liners...?
viva1831@reddit
You'll have to appologise for making it personal, before I respond to demands that I explain myself
sunny0_0@reddit
I'd like to know. Just answer me instead.
sparky8251@reddit
They are lying, thats why they went with "apologize". systemd as an init system literally executes scripts in /etc/init.d as a transition mechanism, and then you can just slap script paths in a service file if you decide to make that too.
I even experienced this transition mechanism as recently as 2 years ago when we finally upgraded from debian 5 systems to ubuntu 20.04 systems and I had to migrate a ton of old custom init scripts...
ECrispy@reddit
Gnome devs are actively user hostile and want to force their own vision on everyone, while removing all possible configuration/settings, dumbing down things, and removing perfectly functional apps with worse new ones. They also don't care much about performance/resource usage/compatibility.
The exact opposite of how KDE works.
Many major distros like Mint/Ubuntu have had to fork Gnome in the past just to provide their users with something usable.
all for some 'its a better user experience' unqualifiable metric, all funded by RH money, and their clout is basically why every distro has defaulted to Gnome and dont offer KDE.
systemd is largely accepted by most of Linux community except some neckbeards who still go on an on about 'one init sysyem' who have no clue what it really is and think its just a startup script. Also there are some valid use cases for non systemd usage such as really tiny footprint distros.
surinameclubcard@reddit
Too monolithic.
lily_34@reddit
I don't hate systemd. I use it, and I like it. But at the same time, I don't want to be locked into it.
degoba@reddit
It was a philosophical dispute. Systemd is not new anymore and it kicks ass
mwyvr@reddit
Ten years since most major distros adopted.
Other systems are capable of kicking ass, but may need a bit more work in certain areas.
Most users could use a systemd or non systemd distribution and never know the difference.
Clearly for that that fall into the most users category, systemd does not provide anything special or necessary but how it does what it does encourages behind the scenes lock down not openness.
How is that a kick ass virtue?
degoba@reddit
Most users. From a sysadmin standpoint its so much better.
Guillaume-Francois@reddit
I don't have enough experience to have any real problem with it, but from where I sit, it does so much that it's going to be an absolute nightmare to replace, which I think would give it an outsized presence in the design considerations of a distro, i.e. you do what the systemd devs say or you go through hell.
I don't have a strong opinion on it, but it strikes me as a bit of a deal with the devil and a potential future nightmare like X was.
mwyvr@reddit
Because Linux is not the only UNIX-like system out there, and because systemd is inextricably linked to Linux and incompatible with operating systems like FreeBSD, OpenBSD, NetBSD and other BSD based derivatives not to mention actual UNIX, Solaris, etc.
UgglanBOB@reddit
Lennart Poettering
paradoxbound@reddit
This the man is an absolute arse. Rude, princess complex. Worse he doesn't understand the needs of his users. I remember one build where he killed all processes owned by the user logging out. Seemed genuinely confused by tools like tmux or running a nohup.
RegisteredJustToSay@reddit
Linux users have a very different culture of adopting change. "If it's not broken, don't fix it" is a very common mantra and mindset, combined with a freedom of choice you don't see in the Windows world. The old system was considered 'good enough' by many and systemd either 'not better enough' or 'more complex' (in a bad way that wouldn't lend itself to easy scripting, etc), so a lot of people got very upset when it started being pushed on them. It felt not just like a deviation from their cherished Linux ideology but also that they were being forced to make the transition against their will.
Very few people actually had meaningful arguments why systemd was inferior - it wasn't, but it caused a lot of tech churn in an ecosystem hallmarked by stability and freedom of choice up to that point.
Prestigious_Pace_108@reddit
Arrogance of some developers.
Narrow_Victory1262@reddit
it tried to make a solution for a problem that not really existed?
Pleasant-Shallot-707@reddit
The problem of enterprise maintainability was real
proton_badger@reddit
Reading the blog entry it's not unreasonable what's being planned. I can see how a DE might ultimately benefit from using more of the functionality being offered by this system management suite. Perhaps things like device management, login management, network connection management and service management.
mwyvr@reddit
Apparently you've not heard of other *nix operating systems like OpenBSD or FreeBSD or NetBSD or Solaris or.. .
Some of these have supported GNOME from the start and still do. An ever deepening dependence om systemd and Linux itself makes GNOME a more closed system and less likely to be supported on non Linux systems.
That isn't a plus.
VictorKorneplod01@reddit
Ok, but how is GNOME being supported by obscure operating systems is a plus for linux users?
mwyvr@reddit
You've missed the point.
tristan957@reddit
Are people from those communities contributing to GNOME? If not, then obviously the DE is not very important there.
LowOwl4312@reddit
I just can't imagine a Unix user deciding to go with Gnome in the first place. Although Solaris uses Gnome, so I'm curious to see what Oracle will do
Business_Reindeer910@reddit
what they will probably do is make their service manage handle these interfaces.
MrHighStreetRoad@reddit
Whether it is a plus or not to the project is what it means for developer output. Maybe there are Gnome developers who are using non systemd systems, who will now leave. On the other hand, gnome developers now avoid having to maintain legacy, redundant code that was apparently holding back some basic features such as session save and restore.
It might not be a plus to other people but it's open source, what contributors do is up to contributors.
proton_badger@reddit
Well, assumptions are assumptions, I started with HP_UX and SUNOS in the nineties.
The obvious issue here that there are benefits in integrating deeper with systemd as described, but at the same time support on UNIX is an issue, as you and the blog mentions. I'm not a GNOME user but I respect their decision to try to push their DE vision forwards with everything Linux is capable of.
In any case the code will need conditional compile directives/runtime checks/compatibility components/etc. as it has already been having for years and like the blog mentions someone needs to continue to maintain that.
It's an unfortunate situation to have to choose direction and lacking devs, but one common for open source. There's not an absolute right or wrong, only trade-offs.
RedditMuzzledNonSimp@reddit
LOL, welcome to your walled garden boys.
mwyvr@reddit
Is gnome contributing to systemd?
AllyTheProtogen@reddit
Honestly, I don't hate SystemD. It's nice, it works, and it's well documented. But making a part of your DE dependent on a certain init system for a higher level of functionality is not a good idea.
ghost103429@reddit
At this point systemD is no longer just an init system but an OS framework. A project for Linux developers to pool resources to create services and tools that would be useful to have when creating and maintaining a distro.
nicothekiller@reddit
Systemd isn't an init system. It's a suite of software. Systemd-init is the init system. It's not a single monolithic giant binary. It's a bunch of different applications under a very similar name.
Existing-Tough-6517@reddit
If v1.1.1 of A requires 1.1.1 of B then A and B are part of the same monolith which is merely factored into multiple executable.
nicothekiller@reddit
Not really how it works. Would be true if you needed every part of systemd, but it's just not the case. You can use a system with systemd-init and literally nothing else. You can use systemd-boot on a system without systemd-init (systemd-boot is literally just another bootloader that got a rebrand). This applies to most of systemd. I know what you are saying, but this simply doesn't apply here. Show me a real example of this in systemd, and I will accept defeat and admit systemd is a monolith.
Left_Security8678@reddit
The reason it got popular its too good. And maintainable
dontquestionmyaction@reddit
And what's your alternative to the problems outlined in the boot post?
This alternative init system offering this stuff does not exist.
Pleasant-Shallot-707@reddit
lol, so make the DE way more complicated than it should be cuz reasons? The distros and the OSs that want to keep using gnome can just implement the headers that Gnome publishes to ensure people can keep using the DE.
MarzipanEven7336@reddit
Hey fucker, you’re copying my comment from Lemmy. But you’re forgiven for spreading the word!
NaheemSays@reddit
The problem isn't that gnome is depending on a specific feature.
It is that the alternate init systems have not implemented it despite there being a need for it going back a decade or more.
How long should desktop environments wait before adopting new technologies?
04_996_C2@reddit
I won't weigh in on the "systemd vs init" debate but I will say that one thing that poisons users against systemd is the adhoc, half-arsed commitment to systemd some distros take. For instance, while I absolutely love Debian, the decision to be a systemd system but not embrace systemd in the networking stack in favor of network/interfaces creates an unnecessary confusion of philosophy of approach. To me it's like the US's approach to healthcare post Obama-Care. You know what's worse than single-pay? A half-assed implementation of single-pay.
Existing-Tough-6517@reddit
I thought the reason not to use this is it sucks
SeeMonkeyDoMonkey@reddit
When the anti-systemd crowd complains that systemd is monolithic...
*Chef's kiss
04_996_C2@reddit
A lot of debates are based on "I feel..." as opposed to "I understand ..."
Existing-Tough-6517@reddit
Distros not wanting to depend on systemd probably also want to skip gnome.
mwyvr@reddit
I don't know, I only manage a few dozen systems. Lots more containers though.
I have systemd based systems and others and I don't see a big difference in benefits or my work effort across all of them.
That comment is related to production systems, not desktops.
There isn't a single systemd feature that makes deploying or administering our desktops easier for me. What am I missing? Anything? Nothing?
This was/is a discussion about gnome, after all.
CmdrCollins@reddit
Gnome is expanding its dependency on systemd to make their lives easier - the same force drove its exclusive adoption by the vast majority of distros, and drives the adoption of other controversial technologies like Wayland.
Jegahan@reddit
That's not entirely true. On one hand yes, one of the reason given is to remove old hacks and replacing them with a more dedicated service thus make the software easier to maintain (which is likely a net positive for users as well in the long term, as less effort is waisted on this, which can free up time for other parts of the DE).
But they also mentioned upcoming features that weren't possible with the old system and that systemd enables, which directly benefits the user:
CmdrCollins@reddit
Choosing to pull in external dependencies over implementing things internally is also a form of making your life easier - there's nothing about systemd that makes it inherently irreplaceable and Gnome could've also chosen the hard way and implemented similar functionality themselves.
Jegahan@reddit
Yes and no. This assumes that they had someone available who is willing to and has the knowledge and time to work on it, while also hoping that it doesn't take to much time away from the other (potentially more important) things this personne was working on.
In pratice, there is a high chance that this option just didn't actually exist and that the choice was between keeping the existing hack and give up on the additional feature, or switch to the systemd alternative.
unknownknown646@reddit
i was actually thinking of switching from MATE to GNOME these days, but no, they had to mess it up. i dont even hate systemd, heck id say its great, and i have absolutely no issues with it, however, FOSS is, well, about freedom, and GNOME here is basically saying "no, you HAVE to use systemd", and i dont think i have to explain the rest.
Jegahan@reddit
Kind of related link : http://islinuxaboutchoice.com/ FOSS is about software freedom. And while this is likely to indirectly lead to more choices/freedom in what to use for the user, this isn't what FOSS is about.
In other words the goal is for software code to be free so that anyone can see, use, modify it, etc as they want. It isn't to provide the user with maximum freedom of software choice or to give them a say in what a person or project does with this code.
I doubt OP read past the headline XD
unknownknown646@reddit
i see, thanks for the enlightment.
pppjurac@reddit
So ... how much functionality of Gnome was removed this time?
Jegahan@reddit
You could just read the article before commenting something wrong, you know?
sudogaeshi@reddit
I can't believe this thread has only 60 comments (for now)
the anti-systemd crowd fading out
MarzipanEven7336@reddit
About god damned time. Whiners need to grow up already, we can’t build good software on top of stinky ass bash scripts.
mwyvr@reddit
Fuck that.
Left_Security8678@reddit
A good thing. Using functionality of 99% of Systems GNOME runs on have just makes sense. Yeah that 1% either have to maintain a Fork or switch Init Systems but eitherway it will severly improve GNOME. I think we should stop to limit ourselves with ideal ideas and face reality as it is, we are the leading FOSS Unix and we shall decide how it should be.
elijuicyjones@reddit
I couldn’t care less what anyone thinks about systemd because 99% of these folks will use whatever everyone else uses and are just pretending to pick a side because they’re so new to Linux they think you’re supposed to.
whitepixe1@reddit
This is bad news for me, I have to ditch Gnome I like because of systemd I despise and not use.
Elbrus-matt@reddit
as a void user,i'll continue to use my wm and xfce4
Isacx123@reddit
Portability out of the window, shame.
NaheemSays@reddit
Not really, the other unit systems just have been exposed as being a decade behind the times and now need to scramble to catch up and add the necessary features.