R.O.S.I.E 5448Mar 14
[object Object]
I’m hoping this doesn’t come off as an “I told you so” post, but a number of the people you dismissed as systemd haters were trying to warn you about the longstanding technical and structural issues that make something on the level of slop code in systemd an instant and catastrophic issue with no easy solution
Mar 13 at 7:31amWeb
Show thread[object Object]Mar 13

are some of the supposed systemd alternatives run by fascists? yes they are. a do-everything init ecosystem is an irresistible lever of power for them. init should be thin and independent enough that it isn’t an ecosystem at all, but if we must have an init ecosystem then we must be very careful who controls that power. again, ideally, there should be no power to control.

with that said, can we please stop fashjacketing all of the people who recognize systemd and wayland as levers of power?

Show thread[object Object]Mar 13

systemd isn’t how most software is shaped. we don’t need to recreate this particular set of mistakes to have a modern init.

“but @zzt what alternatives can you recomm—“ openrc. openrc on alpine or gentoo. openrc is thin enough and reputationally made by people who don’t fucking suck. it’s well-designed. you probably won’t hate it, and the only bits of systemd you’ll miss are the bits there to lock you in.

there’s plenty of other alternatives. don’t post the fash ones or I’ll get fucking mad.

Show thread[object Object]Mar 13
at this point it probably goes without saying but to be perfectly clear: system v init was fucking garbage, as was system v unix. we can do better than sysv or systemd. the cool thing is, we already have!
Show threadd@nny disc@ mc²Mar 13
@zzt system Verybad
Show threadBj�rnMar 13

@hipsterelectron @zzt Anxiously looking for an exit I came across this

> WARNING the instructions are intuitive rather than comprehensive.

at https://wiki.debian.org/OpenRC

OpenRC - Debian Wiki

Show threadd@nny disc@ mc²Mar 13
@zzt openrc is incredibly fucking cool. the maintainer mentioned offhand at fosdem proposing a new shell feature to posix (i cannot remember it offhand but it's one of the basic things bash & co do and i remember desiring it) and everyone in the room cheered and i think maybe everyone in the room cheering about bullying posix is what the future looks like
Show threadwitch_t *naviMar 13
@hipsterelectron @zzt it's arrays

i want proper arrays in shell so that openrc code can be less hacky around storing command lines, and the one per-function array instance (arguments) posix shell has just isn't cutting it
Show threadAnthony SoraceMar 13
@zzt @me_ This is interesting. I’ve been an “anything but systemd” person (seriously, just an init script is better in >50% of cases) since it showed up, but I’ve somehow missed some having fash associations. Is there something you can point me at to read so I can learn more?
Show threadArthfach 🐻Mar 13
@zzt Good to know. I've been looking at init systems and OpenRC definitely was appealing; this confirms my suspicions.

The real trick now will be figuring out how in the world I convert my production server (Debian 13) from systemd to openrc with minimal downtime and without a reinstall.
Show threadKrishean DraconisMar 13
@arthfach @zzt if you figure this out please document it, i'm sure there are many who would be interested (including me)
Show threadLuna LacteaMar 13
@arthfach @zzt I thought switching init systems was the one thing that sold never be done without a completed reinstall, just because of the amount of work it takes & that it could make your system very difficult to work with later. I don't know much about init systems though. Is it really that easy? Is there really an option for Debian with something other than systemd (that isn't Devuan)?
Show threadKirtai 🏳️‍⚧️Mar 13

@zzt
There's even Shepherd.

(The Guix team have been quite noticeably pulling away from the FSF for a while now)

Show threadKirbyMar 14
@zzt what fash ones? and why would i care if you get mad?
Show thread[object Object]Mar 14
@kirby what a fucking reply
Show threadlizzy مثلية الله ☭  Mar 13
@zzt which ones are run by facists, huh? i know openrc, s6 and runit and nitro are done by cool people. is dinit bad or smth?
Show thread[object Object]Mar 13

@lizzy the best examples I have of the ones reaching for an ecosystem are fortunately dead systemd forks run by absolute assholes.

the ones I’ve seen pointed to as examples in modern discourse are suckless init, which I don’t think anyone uses (for good reason, it’s baby’s first init made by nazis), and devuan, which doesn’t actually run an init project of its own but which is the most visible fascist anti-systemd ecosystem.

Show thread[object Object]Mar 13
@lizzy usually suckless init and devuan are synthesized together to form the incorrect accusation that systemd alternatives are all run by fascists. it’s a ridiculous thing that falls apart under any analysis, but unfortunately that’s where some of the discourse has been lately.
Show threadlizzy مثلية الله ☭  Mar 13
@zzt oh yea, artix unfortunately is full of fash people too

i used to run artix for years because my laptop wasn't powerful enough for gentoo, but now gentoo has the binhost and i have better hardware, so i was able to switch back. really glad i did considering they package xlibre now
Show thread[object Object]Mar 13
@lizzy uggh xlibre is such a prime recent example of attempted ecosystem control too. it’s not a competent attempt on a technical level at all, but it gets marketed and SEOed to all hell in an attempt to astroturf legitimacy. the hope, I think, is for people to eventually run out of alternatives and get desperate enough they’re willing to look past the fascism, and then they’re stuck.
Show threadlizzy مثلية الله ☭  Mar 13
@zzt i still love how the xlibre people wrote 2^16 in C to mean 2 raised to the power of 16
Show threadnickMar 14
@lizzy @zzt Making this even funnier, the gcc warnings for this are just so good and helpful:
Show threadAndrea (Drea) Tamar PinskiMar 14
@nickzoic @lizzy @zzt I have to give credit where credit is due. It came originally from clang. GCC bug report mentioning that:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90885#c20
90885 – GCC should warn about 2^16 and 2^32 and 2^64 [-Wxor-used-as-pow]

Show threadMoses IzumiMar 13
@lizzy @zzt
>binhost
You're telling me Gentoo officially lets me fall back on binary packages?
Sick!
(guess I have my ticket away from the Debian world sorted then)
Show threadJacob Alexander TiceMar 14
I'm really in an annoying pickle with that because I'm just going to be honest, I don't like Gentoo very much. I am most comfortable with Arch Linux's tooling. I am more comfortable with Pacman than literally any other package manager. I don't care if Portage is technically superior or more powerful or whatever else, Pacman and makepkg are simply how my brain works and there is no user repository that I prefer pulling packages from more than the AUR. But, like, Artix has some rather dubious people involved in the project and Obarun is run by promptfondlers. So, like, what's my option here?
Show threadlizzy مثلية الله ☭  Mar 14
@daemonspudguy @zzt I think you can also install openrc on arch, from AUR... and then also install service files for various things from AUR, but i think at this point it's actually easier to use Artix
Show threadJacob Alexander TiceMar 14
Honestly I'm probably just going to do that. At least it still packages Xorg so I can avoid XLibre like the plague because it is the plague.
Show threadLucasMar 13
@zzt @lizzy What's a bit frightening is that by a quick glance on Devuan's website + Wikipedia, I never would have guessed that there's anything going on. It's also hard to find sources that go into more details (search brought up a podcast by ugh.... Lunduke that I'm *not* going to listen to 😒).
Show threadφMar 13
@zzt where does wayland come into this at all ??? how is that a level of power ?? the point of wayland makes it even less centrally controlled than x11 was , witj pretty much only one working implementation . i agree with the sentiment of this posts but wayland has many competing implementations both toolkitwise and composotor(library) side , it does NOT suffer from this
Show threadφMar 13
@zzt not liking wayland and freedesktop is fair but like . lets not pretend your point wouldnt stand even stronger in an xorg controlled situation , please
Show threadφMar 13
@zzt wayland puts more power in the hands of developers
Show threadφMar 13
@zzt whicj is also a fair criticism of wayland : “i dont want to implement so much stuff just for what would be a simple window manager under x” is a fair argument !!
Show threadkopper Mar 14
@fiore @zzt to be fair, that's not the protocol's fault per se. a compositor can decide to outsource the window management bits to a separate process. in fact, it already exists codeberg.org/river/river
river

A non-monolithic Wayland compositor

Codeberg.org
Show threadφMar 14
@kopper @zzt indeed , i love river ! the point is , there is more choice on “who has power” when it comes to wayland
Show threadjwzMar 13
@zzt I probably really don't want to kick this hornet's nest, but ... I hated init.d and I hate systemd at about the same level of hate, how is one of them trying to take my freedoms or whatever? I missed the memo. (Seriously I'm not being sarcastic I missed the memo)
Show thread[object Object]Mar 13

@jwz that’s fine, they’re both trash

keep an eye on your timeline for posts from within the last 12 hours or so lamenting how much of the critical Linux userland infrastructure under systemd just became slop for a worked answer to your question. we’re unfortunately past the point where the damage is theoretical.

Show threadspinnyspinlockMar 13
@zzt You can say I told you so, as a treat
Show threadsarah tonin Mar 13

@zzt honestly it doesn't even feel like a gotcha or "i told you so" moment, people got tricked into abandoning redundancy for their systems, and now are left to pick up the pieces with the guilt that they helped push something that ultimately ended in a disaster.

it's all just sad

Show threadbeanMar 13
@SRAZKVT @zzt i do still kind of like systemd, and if there was a maintained fork i might try it again in the future. but slop is a pretty big red line and i'm excited to try the alternatives and see how wrong i was :3
Show threadsarah tonin Mar 13

@uproot4269 @zzt i'll be extremely honest, the chance of a succesful systemd fork is close to 0. project forks only work when you have people who already at least somewhat know how the project works, preferably existing recurrant contributors (because if you did one contribution 5 years ago that fixes a typo somewhere, you don't know anything about the project's concrete structure) and maintainers, but if the maintainers themselves are pushing for things harming everyone else, they're not going to be on board.

also, the very architecture and design of systemd, aka, taking every component and putting them, tightly coupled, into a single mostly homogenic slab of code is exactly the reason why we got into this mess, a fork will not fix that