[ home / rules / faq / search ] [ overboard / sfw / alt ] [ leftypol / edu / labor / siberia / lgbt / latam / hobby / tech / games / anime / music / draw / AKM / ufo / 420 ] [ meta ] [ wiki / shop / tv / tiktok / twitter / patreon ] [ GET / ref / marx / booru ]

/tech/ - Technology

"Technology reveals the active relation of man to nature" - Karl Marx
Name
Options
Subject
Comment
Flag
File
Embed
Password(For file deletion.)

Check out our new store at shop.leftypol.org!


File: 1793296802080-6.png (238.98 KB, 474x296, ClipboardImage.png)

 

I know this might be something thats asked a lot but, what do you guys use/reccomend in terms of linux? I have been in to linux and other stuff for a while, distro hopping from debian to ubuntu(and its variants) to finally cachyOS/void system in my computers. Should i just fully switch to void or just install arch in itself? I was thinking about installing Deepin, and i think i am going to try that one.

>linux
vibecoded slop
I'm considering switching to netbsd

>>34699
Honestly I tried NetBSD before but the experience isnt really different from linux, switching to BSDs just means less support for stuff. Things like DEs and most common software are the same anyways.

I still like it though, it's pretty good if you still need to use an old computer daily.

I've been a fan of Gentoo for many years now. Mostly just because of how easy it is to patch and create ebuilds in a local repository or using the patch folder. Reducing runtime dependencies like DBus is also nice. The few init services I've written have not been so difficult.

I wouldn't recommend it for everyone though, it depends on your needs, and the resources you have available. Also don't claim to know everything about system administration. Want binary package for my server - but Gentoo has these too, for amd64 at least.

Only other thing I'd really consider running atm is Devuan.

Alpine no question. While it's what everyone uses for their lightweight containers and VPSs, for anyone cognizant of the system it's a fine desktop as well. If you need to run any glibc binaries, just be prepared to install gcompat or a chroot.
>>34699
The parts that matter are shared among all *nixen anyways. It's extremely unlikely LLM reviews will cause a glaring bug to slip into Linux (the kernel), yet it already happened with rsync and no doubt will also affect the pkgsrc ecosystem as a whole.

>>34702
>LLM reviews
that was two months ago, now they are fully vibecoding it

>The parts that matter are shared

literally not the case. the codebases are completely different, the api has superficial similarities but it is also different

>>34727
>the codebases are completely different
I was talking about third-party software aka. pkgsrc.

Fedora just works so it's what I prefer. If I were to switch, I'd go to Arch. I generally dislike forks of distros, just can't trust them. I install Linux Mint if it's a computer for someone else. Debian just isn't great for desktop use in my opinion, everything is outdated and I've had problems because of that. Ubuntu nowadays is out of question, I'd rather run Windows if it came down to that.

File: 1794129444065-7.mp4 (102.42 KB, 468x58, l3atdjs69opg1.mp4)

KDE Linux if you just want something that works out of box & is stable and you have no computer literacy

Fedora Linux for the general user/gamer

GNU Guix for the software freedom enjoyer who can afford a modern laptop that can boot the Linux-libre kernel (normal Linux kernel can be used through the nonguix channel though, so you can still use it on a normal PC)

Avoid Alpine/BSD operating systems entirely, MIT/BSD is a fascist license and you should aspire to stick as close as possible to FSF/GNU even if it requires using lower-quality code

Avoid Linux Mint and anything which uses the X Server instead of Wayland
Avoid Ubuntu due to non-free packaging system (Snap), unless you're a musician, in which case Ubuntu Studio is your best bet at Linux anyways

Stick to upstream distributions as much as possible, Arch, Debian, openSUSE, Slackware, Fedora, Gentoo, etc. will save you a ton of headaches and are much better for newbies as they do not do things differently (EndeavourOS and Manjaro are two examples of distributions that change things that don't need to be changed without proper documentation, making them buggy in ways that your average computer user would never be able to know how to fix)

>>34762
elaboration: by Alpine I mean any distribution of Linux which carries musl/BusyBox instead of the GNU Operating System utilities

also, "vibecode slop wahhh" is a meaningless thing, every widely used open-source operating system thus far has been funded by literal rapehordes, companies like Dell and Red Hat that are known to directly profit from genocide, this has been happening the past 20 years, tech is an integral part of the military industrial complex, using another bourgeoisie tool makes fuckall difference in ethics

as for any sort of cope about the code quality, Linus Torvalds checks every AI-generated commit before it is merged, one of the finest software engineers currently alive, Linux is good because of him.

File: 1794140337238-4.jpg (132.96 KB, 1085x1507, smirk_sweater_cup.jpg)

>>34762
Eschewing higher-quality software in favor of the FSF's nebulous concept of "software freedom" is actually existing idealism.
>b-but it's le gift economy
Only in the sense same sense that hobbyist MIT/BSDware is. Aside from legacy vendors, software companies no longer need licenses to project ownership onto a project, Android being the most prolific example. The Linux Foundation has long been the embodiment of a corporate consortium, while the FSF operates differently only by virtue of being run by hobbyists, which is equally true for something like NetBSD.
>>34763
>musl/BusyBox
Musl has been a one man project for most of its lifetime and you do realise busybox has the same license as linux, right? To cast aside the notion of this having anything to do with the GPLv2, consider the matrix.org webclient, which is solely licensed under the AGPL and as corporate-driven as it gets. Lastly RMS himself has argued in favor of these types of projects with his "not free as in beer" rhetoric, pointedly ignoring the fact, that having the source to Windows XP for example is only of marginal use in most scenarios.
>Linus Torvalds checks every AI-generated commit before it is merged
This shit has been shown to actively rot your brain, so enjoy it while it lasts. See the recent rsync regressions.

>>34765
"Eschewing higher-quality software in favor of the FSF's nebulous concept of "software freedom" is actually existing idealism."
It's not idealism because it is actually existing, software already works on glibc and GNU, all professional FOSS and games work on GNU, but not BSD userlands or plan9.
we already have a working standard that ensures user freedom, using anything else and contributing to it is fascist fragmentation
"you do realise busybox has the same license as linux, right?" not a good thing by the way, GPLv2 is simply the best we have right now, Hurd will be stable in 2-3 years though so I won't have to care anymore.
"This shit has been shown to actively rot your brain, so enjoy it while it lasts. See the recent rsync regressions."
noone above 50 and is a distinguished compscientist/engineer is going to have their brain "rotted", years of school and work would have already built your attention span.

Also, pray tell, how are you going to achieve any sort of cybersecurity if LLMs can find exploits at x1000 the speed you can patch them without using the same exact LLMs? the cat's out of the bag already
"corporate-driven"
corporate-driven =/= infringing on user freedom
"software companies no longer need licenses to project ownership onto a project,"
what? yes they do lol, the only reason Android got like that is because of Apache. Android is the operating system and Linux is the kernel.
this is why - a rational person makes sure that their operating system contains as much GPL-licensed code as physically possible.
If you statically or dynamically link your proprietary code directly against a GPLv3-licensed library or component, your application generally becomes a derivative work and must also be licensed under the GPLv3.

File: 1794160979544-9.jpg (1.41 MB, 1265x1800, casual_sit_lean.jpg)

>>34766
>corporate-driven =/= infringing on user freedom
This is where your argument in favor of the GPL falls apart. It should be obvious to anyone, that proprietary software licenses are a side-effect of the profit motive overriding actual utility, not itself the impetus of "infringing on consoomer rights". If the GPL doesn't actually prevent suits from profiting off of volunteer work or making their own user-hostile changes, any freedom it affords over other source licenses is purely theoretical (read: terminal liberalism). The original qmail license would genuinely be better suited for that, because it scares the hoes.
>muh fragmentation
Not having a linux/gnu/freedesktop monoculture is good actually. For one the GNU userland would be an active hindrance in certain applications like embedded and portable standards are generally good for software ecosystems, because they hold code to a higher metric (shitshows like literal Eunice and Apple's slowly collapsing C/UNIX interface exempted). If you had ever been hit by any glibc or linux-isms (sometimes forward-incompatible with linux/glibc itself) while compiling production software, you would probably understand.
>noone above 50 and is a distinguished compscientist/engineer is going to have their brain "rotted"
What was shown, is that working with an LLM can actively make you worse at whatever task you're trying to accomplish, regardless of prior training. At the same time people will feel like they're more productive, because of the constant self-validating feedback loop. At the end of the process it won't matter if they have the training to do it right, because they will no longer want to.

>>34768
"If the GPL doesn't actually prevent suits from profiting off of volunteer work" it does, why make shit up? Just to be retarded??
"For one the GNU userland would be an active hindrance in certain applications like embedded and portable standards are generally good for software ecosystems" entirely useless and meaningless in the context of user desktops, the enterprise and industrial people can use whatever they like.

Anyways, you're pulling up a lot of retarded references and talking about things nobody does (like mentioning qmail, Eunice, "compiling software" as if anyone on earth actually does that shit) - given the smug Cirno pictures too and luddite arguments of "calculators fry your brain" I'm guessing you're another Unixtard and not someone actually concerned with combatting the surveillance state

File: 1794206224131-2.jpg (938.92 KB, 910x1321, maid_smug.jpg)

>>34769
>"compiling software" as if anyone on earth actually does that shit
So you're not even making use of most of the "software freedoms" and yet you have the gall to insult the work of Rich Felker.
>enterprise and industrial people can use whatever they like
RMS started his crusade over fucking printer firmware, so yes it is beneficial to the user, that they can just telnet into many embedded devices and be greeted by good-old busybox.

Your reason for using GPLed software seems to boil down to radlib virtue signaling, because it's entirely harmonized with technocapital (the actual "surveillance state") and you can't even materially back up why it differs from MIT/BSDware from the user's perspective. I'm done arguing with you software illiterate shit.

>>34771
"So you're not even making use of most of the "software freedoms" and yet you have the gall to insult the work of Rich Felker."
Whether I compile my software or not doesn't make me "make use of software freedom" less, Rich Felker is a POSIX retard, as POSIX is a retarded standard which not even the BSD systems can comply with anymore.
Again, autistic compsci stuff that eliminates user freedom by advocating for MIT-fascist software for some sort of principled Unix Plan9 Philosophy only nerds care about (makes 0 difference to the user, by the way)

"so yes it is beneficial to the user, that they can just telnet into many embedded devices and be greeted by good-old busybox."
telnet into the Microsoft Telnet Client, right?


"Your reason for using GPLed software seems to boil down to radlib virtue signaling, because it's entirely harmonized with technocapital (the actual "surveillance state") and you can't even materially back up why it differs from MIT/BSDware from the user's perspective."
Search up 'tivoization', "entirely harmonized with technocapital" yet only GPLv3 software could prohibit the problems that led to https://keepandroidopen.org/
Huh. Funny how that works.

"I'm done arguing with you software illiterate shit."
Good, fuck software literacy and the legacy fascist dogshit brought on by MIT and Berkeley, proletarianize software.

>>34763
>Linus Torvalds checks every AI-generated commit before it is merged
Do you know how the Linux project even functions? Nowadays commits are first reviewed by a shitty system known as "Sashiko" that's basically just an LLM wrapper made by Google and which has been noted to actually biased in favour of LLM commits. This is then sent to a team of maintainers, who despite being a group are already overwhelmed by the slop commit flood.

Linus doesn't do shit personally, and maybe that's for the better since his own big slop commit last month contained a race condition that was spotted by a human.

>one of the finest software engineers currently alive

Lol. The big slop commit I mentioned above was made with Python and Linus mentioned how cool it was that LLMs now make him capable to make Python commits since he always struggled with it (an abstracted high-level language taught to high schoolers).

>>34773
Point incomprehensible, do you think we didn't have code review automation software before LLMs? or is your point somehow invalidating Torvalds' well-proven track record of the past two decades by the fact he made 1 error?
Why would he not struggle with an operating system he was never trained in?

>>34776
sorry, programming language

>>34776
>Point incomprehensible
The point is that
<Linus Torvalds checks every AI-generated commit before it is merged
is just plain wrong.

>somehow invalidating Torvalds' well-proven track record of the past two decades

His technical track record consists of being a medicore C programmer (which is probably not true anymore since he let his skills rust lol), and his non-technical managerial track record consists of constantly fellating corpos, jamming their garbage into the project or kernel itself, and keeping the GPLv2 specifically to protect his corpo patronages and spite Stallman.
I don't even know why someone who behaves like the FSF's strongest soldier has personality cult around this guy since he's very obviously opposed to the whole thing.

ngl the vibe coded shit is going to make me uninstall arch in favor of an LTE distro, it's not even limited to linux like the other guy said, but the entire toolset is already turning to vibe slop. it wouldn't be that bad, if the corollary wasn't the expectation of shipping faster which is going to break something eventually, and if something breaks, rolling release distros are going to be the first to feel it.

>>34762
>(EndeavourOS and Manjaro are two examples of distributions that change things that don't need to be changed without proper documentation, making them buggy in ways that your average computer user would never be able to know how to fix)
ive heard that about manjaro. afiak the only think endeavour does is graphical install with choice de, preinstall barebones basics, yay for pacman, and dracut over mkinitcpio.

im bias, cause i use it, and imma install arch if i ever have to reinstall but i wonder what else they do. it was the highest arch based on distrowatch back when i chose, now cachy is but im not a gamer and think its bloated.

>>34784
An LTE distro is just going to make it worse- now you're sticking old vibecoded software instead of fixed vibecoded software

>>34786
"now cachy is but im not a gamer and think its bloated."
You can't call CachyOS bloated while using EndeavourOS/Manjaro
Either way, EndeavourOS had weird changes assigned to their X server on some of their DE install options, this was years ago but it was just genuinely really fucking stupid
No point in them, stick to upstream.

>>34773
so basically the linux kernel is being developed by a google bot? great.

>>34788
No grasp on what 'code review' means?

>>34788
No it's that every change must now be reviewed by an LLM first.

The funny thing about it is that it gets gamed and basically prompt injected by LLM commits, which are almost always accepted even if they're a +10k diff mess. Meanwhile human commits get nitpicked by it constantly.
What was originally intended to filter out the slop so that the maintainer reviewers don't get overwhelmed did the opposite of that.

>>34787
>You can't call CachyOS bloated while using EndeavourOS/Manjaro
i dont really think you can say that if you equate those two

>>34798
What makes them different? That CachyOS precompiles it's packages or that it has multiple kernels on install? They're both effectively identical in bloat otherwise.

>>34804
> or that it has multiple kernels on install?
i think so, yeah. it has a dedicated "gaming" kernel with its own in-house scheduler supposedly, it also has a kernel for handhelds. i have no idea if it makes any difference, i installed it because the bazzite installer wouldnt work on a gpd win 5


Unique IPs: 21

[Return][Go to top] [Catalog] | [Home][Post a Reply]
Delete Post [ ]
[ home / rules / faq / search ] [ overboard / sfw / alt ] [ leftypol / edu / labor / siberia / lgbt / latam / hobby / tech / games / anime / music / draw / AKM / ufo / 420 ] [ meta ] [ wiki / shop / tv / tiktok / twitter / patreon ] [ GET / ref / marx / booru ]