Login
You're viewing the functional.cafe public feed.
  • Oct 16, 2026, 8:39 AM

    Nothing new, but I'm disappointed.

    Many years ago, Linux represented the alternative. The question was: "Why don't you just use Windows? It's easier." And the answer was, almost always: "To be free."

    The "bazaar" spirit of Linux and its distributions made us feel free. Personally, I've always thought that the "cathedral" approach of the BSDs produces more reliable and maintainable results. But that feeling of freedom was fantastic.

    There wasn't a company deciding everything. There was a kernel developed by people following what they believed was right. Or what the community, in one way or another, was asking for.

    Torvalds has always been quite outspoken (remember the middle finger to Nvidia?). But the general spirit was, again, freedom. Anyone, sitting in their bedroom, could contribute in some way. Even me.

    Today, after the embrace of AI, I read statements about wanting only "serious" contributors, "serious" contributions, because the world depends on this code and reliability must come first.

    I understand the problem. Maintaining the Linux kernel is a huge responsibility. And low-quality contributions, especially those blindly generated with AI, are a real problem.

    But what I hear, between the lines, is something like: "Hey, kid in your bedroom. If you don't have the right tools, the money to afford them, or the experience we expect, maybe your contribution isn't even worth considering."

    Am I exaggerating? Maybe. But this is what I got from the tone and content of the latest public statements.

    And it makes me sad.

    Because it feels like a transformation towards what we wanted to escape, back then.

    So please don't ask me why the BSD community, imperfect as it is, feels like home to me.

    When I see young people, with limited resources, being able to contribute and attend conferences (thanks to the support of the Foundations), and I see the enthusiasm in their eyes, I feel at home.

    Because I still remember what it meant to be that kid in a bedroom.

    And I really hope we never close the door in their face.

    #IT #ModernWorld

    💬 27🔄 163⭐ 229

Replies

  • Oct 16, 2026, 8:56 AM

    @stefano amen (and I don't think you're exaggerating; the prices of the cheapest AI subscriptions exclude most of the planet already, and local models are outlandishly expensive except maybe for the 1%, while having extremely poor quality!)

    💬 0🔄 5⭐ 15
  • Oct 16, 2026, 8:57 AM

    @stefano I partly agree with your feelings.
    On the other hand, once a project gains a "critical mass", sure you don't want to break it by flooding the gates with low-quality contributions either.

    Back then was Linux still small and more an hobby project that a real production one. But once the TOP500 has been made only by servers running Linux, you know it's not a pet project anymore but it is "serious", while the stakes are bigger and for sure there are player pouring money into it.

    💬 2🔄 0⭐ 3
  • 💬 1🔄 3⭐ 8
  • 💬 1🔄 0⭐ 0
  • Oct 16, 2026, 9:45 PM

    @madduci @stefano Yea, there was an excellent thread here on Fedi recently, which I don't have a link to rn. But the crux was "Linux needs to be honest about whose needs have priority, b/c priority exists whether they acknowledge it or not".

    💬 1🔄 0⭐ 0
  • Oct 17, 2026, 6:03 AM

    @cr1901 @stefano I personally don't blame it, since open source projects have always been struggling with funding and obviously, once a project get good fundings, it would be counterintuitive to not support who gives you money

    💬 0🔄 0⭐ 0
  • Oct 18, 2026, 4:58 PM

    @madduci @stefano OTOH, to prevent low quality contributions from young developers there have always been code reviews, hopefully done in a way that allowed motivated people to learn what they were doing wrong.

    they have always been a cost on the project, but also hopefully an investment, as a way to get new developers.

    Of course, low quality contributions from AI using people is going to be 99% cost and 1% investment, and that's in the hopeful assumption that 1% of those may start at the very least reading and fixing what they are submitting before doing so, breaking this model.

    💬 0🔄 1⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 2
  • Oct 16, 2026, 9:15 AM

    @stefano "When I see young people.." <- what conferences are you attending because I haven't seen much or any of that in the last 10y or so. When our generations go that's gonna be it. We're gonna be maintaining shit in our 70s, like the lords of COBOL. Sorry for such pessimism early in the day.

    💬 3🔄 0⭐ 6
  • Oct 16, 2026, 9:17 AM

    @Nux not the majority, but at EuroBSDCon there are younger (< 30 years old) young people. But you need to travel, pay for your ticket. I'm not sure I could afford that in my 20ies. But some of them make it, and the foundations are giving some funds to help them to contribute and reach the conferences.

    💬 2🔄 0⭐ 5
  • 💬 0🔄 0⭐ 3
  • Oct 16, 2026, 5:00 PM
    @stefano @Nux yeah - I want to go to a basic security conference - the one near me that I can get to on the train. BUt the tickets are $1500. So many large cons are for *employees* not learning.
    💬 1🔄 0⭐ 0
  • Oct 18, 2026, 5:37 PM

    @silverwizard @Nux @stefano yeah, in my experience there are two very different things under the name “conference”

    one includes community conferences, hacker camps and the like, is usually free or almost free to attend (with voluntary tickets, and of course there may be optional accomodation to pay, usually of the cheaper style): these are designed for people to learn and work together.

    the other one is designed to facilitate networking for people with already good salaries and companies, and is pretty much unrelated for the former, and comes with astronomical ticket costs.

    (and then there is the third one, academic conferences, with their own peculiarities)

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Oct 16, 2026, 9:28 AM

    @stefano

    Over 30 years of the Linux Desktop at this house, but this morning has begun a new era of the NetBSD Desktop. The install on my newish laptop is a success! The happiness that began then, is once again now. I feel 30 years younger 😹

    💬 1🔄 0⭐ 14
  • 💬 0🔄 0⭐ 1
  • Oct 16, 2026, 9:58 AM

    @emilianosandri @stefano

    This is a Thinkpad P15 Gen1 with the I9 processor, htop is showing 16 cores, lspci is showing both Intel CometLake and nVidia Quadro T2000. Framebuffer is a bit wonky with the buffering, but absolutely usable.

    💬 0🔄 0⭐ 3
  • Oct 16, 2026, 9:36 AM

    @stefano I wouldn't be concerned too much for #Linux . The project has gone through much worse, but it's possible for this AI drama to end in a rollback or a fork war.

    Linus is clearly going through a midlife crisis ATM. I don't blame him, he's never been perfectly stable, and this AI crap is driving many of us insane already..

    This #AI stuff will likely calm down once the bubble bursts, AI companies backtrack/go broke, AI becomes unaffordable, and Linus is forced to touch grass once again.

    💬 1🔄 3⭐ 11
  • Oct 16, 2026, 9:43 AM

    @alice_pea_3526 Probably. But the point isn't about AI. It's how the spirit of the project is changing. From the "we're open" to "if you're not good enough (and good can mean a lot of things), you're out".

    Sure, they need to filter out. The project is now too big to accept any kind of contribution. But sounds like... some sort of... well, I can't find the right word, but I don't totally like it.

    💬 1🔄 0⭐ 3
  • Oct 16, 2026, 7:18 PM

    @stefano @alice_pea_3526 The Kernel team drag themselves in this situation taking this open door approach, nevertheless. I'm not impressed at all they now lament to get flooded.
    Other FOSS projects already took precautions for avoid this.

    Frankly, sometimes I think the Linux Kernel was from long time the place where too much people poke their hand into, for whatever good and bad reasons, so I'm not surprised they are so open to get flooded by AI stuff.

    💬 0🔄 0⭐ 0
  • Oct 16, 2026, 10:02 AM

    @stefano Maybe it's just age – Linux has grown up and is old. Computing is commoditised and boring.

    If I were a kid today, I don't think I would hack the linux kernel, or Wayland.

    I would probably hack around with physical stuff, 3D printing, drones, robots, smart gardening, whatever.

    There would be a ESP32 or Raspberry somewhere, in the same way there was electronics somewhere in my Commodore 64, but that would not be the important bit.

    wiesmann.codiferes.net/wordpre

    💬 1🔄 0⭐ 5
  • 💬 1🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 1🔄 1⭐ 4
  • 💬 1🔄 0⭐ 0
  • Oct 18, 2026, 1:12 AM

    @mewsleah

    NetBSD does have famously extensive architecture support, but aside from GPU support Haiku may be better for certain older and newer PC hardware combinations. Haiku also reuses most networking drivers from both FreeBSD and OpenBSD, which gives it a rather broad Wi-Fi adapter support in particular (except for the bwl driver most notably).

    As is usual with OSes that aren't mainstream, hardware support can be spotty, requiring research and actually trying on bare metal to know how well your hardware works with them.

    Beyond hardware compatibility, Redox and Haiku may also be easier for some people to get going with since they have graphical desktops by default. In Haiku's case, it's only graphical and built around that design decision.

    💬 0🔄 0⭐ 0
  • Oct 16, 2026, 10:37 AM

    @stefano @viq I remember a time before Linux existed, when most of the Unix variants in the world were proprietary and based on BSD. When SCO was a cool upstart company to work for and 386/BSD struck terror into the hearts of commercial Unix vendors.

    Fading empires with entrenched interests are always going to take irrational actions that ultimately end up destroying them. The Unix vendors of the time stopped innovating and started litigating. Innovators routed around them and eventually commodity CPUs and the OSes running on them became “good enough”. It was an extinction level event for the proprietary Unix vendors who had seemed untouchable.

    I didn’t get into computing because of an operating system. I got into computing because I wanted to understand how everything worked. I gravitated to Unix-like operating systems because they seemed to give me the most tools to accomplish that mission. I learned to code to test and prove my understanding.

    And I’m still learning, still testing, still proving, and still improving myself. Keep focused on your mission, whatever it may be. Fight the battles that need fighting. Remember history has shown the things that seemed most unchangeable have a habit of suddenly becoming something remembered only by old timers like me.

    💬 2🔄 7⭐ 15
  • Oct 16, 2026, 2:21 PM

    @hal_pomeranz @stefano @viq

    +1

    lived through most of that unix history too. i also remember all the various networking protocols, which wound up with no proprietary cruft, just TCP/IP.

    good history to learn in both cases. vendors trying to lock you in, with no concern for anything but their profits, don't do as well as products that solve your problem but are open/interoperable.

    💬 0🔄 0⭐ 0
  • Oct 17, 2026, 9:27 AM

    @hal_pomeranz @stefano @viq

    What I'm sensing is that it seems we're returning to a new cycle similar to the early '90s, where distros that are now commercial and backed by corporations (Ubuntu, Red Hat, etc.) are the new UNIX operating systems (HP, etc.) of that era, and the other distros are striving to stand out and remain the rebels, outside corporate logic.

    The difference, though, is that back then, people tried to innovate and stand out in areas where the big UNIX players could no longer do so; today, unfortunately, decisions made by the major distros are adopted in a chain reaction by the others (see systemd or Wayland, etc.), and as a result, anyone who doesn’t follow suit ends up “left out.” I think people are talking about and taking an increasing interest in BSDs (and thank goodness, I say) after years in the shadows, because they see them as alternatives to the major distros and Linux in general—just as Linux was to the big UNIX systems back in the day (not to mention the freedom it offered by not having to use Windows).

    💬 0🔄 1⭐ 3
  • Oct 16, 2026, 10:38 AM

    @stefano AI(LLM companies) have double-crossed the community twice: on the one hand, used open forums for training, including those that explicitly prohibited training, while, on the other hand, it’s automating the community away. This destroys opportunities for exchange and learning, and thus destroys the community. My theory is that there will come a point where Linux itself will be replaced by AI. What will happen then?

    💬 2🔄 1⭐ 6
  • Oct 16, 2026, 11:22 AM

    @thec @stefano What do you mean replacing Linux with AI?

    They are not the same thing and I don't see "an AI" (whatever that is by your definition) run a computer as long as we have the architectures we have.

    💬 1🔄 0⭐ 0
  • Oct 16, 2026, 11:26 AM

    @ftranschel @stefano The term (AI) is actually intentionally kept a bit vague, but right now I primarily talk in the context of LLMs and partly the companies behind.
    My guess would simply be a “white room” implementation (which might have been a little bit too vague, sorry). In addition, there have been various attempts to build “LLM-first” systems and also the possibility of different architectures. The question is simply how this will develop from here.

    💬 1🔄 0⭐ 0
  • Oct 16, 2026, 12:37 PM

    @thec @stefano Ok so if I get you correctly you suggest that #Linux will be made irrelevant by an #AI #kernel, yes?

    I think experiments from #anthropic showing that you can #cleanroom a #compiler to *some* extent (pathetic as it may be) clearly show that it probably could be done, but the question of course is how to streamline it in a way that #Linux provides right now for a lot of architectures. I think this would #tokenmaxx quite quickly...

    💬 1🔄 0⭐ 0
  • Oct 16, 2026, 1:45 PM

    @ftranschel @stefano that is true for now. In the long run, I still wouldn't bet my life on it. However, that's also why it's just a theory.
    It might be worth adding that my point also isn't with the use of LLMs in general, but rather with the treatment of the community specifically.

    💬 1🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Oct 16, 2026, 7:23 PM

    @thec @stefano maybe soon or later Red Hat is going to "fork" the linux Kernel for "better integration with SystemD"...
    :trollface:

    💬 0🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • Oct 16, 2026, 7:44 PM

    @stefano @thec there is this long standing joke about SystemD replacing the Linux Kernel, but I feel somehow this isn't going to stay a joke somewhere in future.

    💬 0🔄 2⭐ 0
  • Oct 16, 2026, 11:29 AM

    @stefano It's not just the experience or the tools, it's also "how aligned are you to MY management?".
    The freedom part, for those so called open source projects, is just a collateral historical blabbing

    💬 0🔄 1⭐ 5
  • Oct 16, 2026, 12:02 PM

    @stefano Where you go wrong is that if said kid in the bedroom contributes something and has no clue what that contribution is, does or what its impact is, it really isn't worth considering. Irrespectively whether that contribution was generated by AI or not.

    And the experience by most open source project owners is that most of the AI generated contributions are slop, can't be explained by their submitter, and take up huge amounts of time from the projects maintainers.

    💬 1🔄 0⭐ 3
  • 💬 0🔄 0⭐ 1
  • 💬 1🔄 0⭐ 1
  • 💬 1🔄 0⭐ 0
  • Oct 16, 2026, 7:28 PM

    @stefano @FossThought both have a lot of potential as kernel, really. I do not deal with software at that level, but as much I understand, the anykernel approach of NetBSD is what an other project lacks and could fit better for the needs of today, where portability is getting critical with the advancement of other platforms like Arm and RiscV.

    I have a lot of hopes for Hurd, that would solve a lot of hurdles that are genetic to Linux and gives some headaches for desktop (aka drivers!).

    💬 1🔄 0⭐ 0
  • Oct 16, 2026, 7:31 PM

    @raster @FossThought I've been reading great things about Hurd since the late '90ies. It's still a WIP. I hope to see it somehow being usable in the proper way before I'll be in my 90ies.

    💬 2🔄 0⭐ 0
  • Oct 16, 2026, 7:43 PM

    @stefano @FossThought maybe in the next century. :guilded_lul:

    But seriously, finally having a working version on 64bit machines (thanks also for the code transfusion from NetBSD, btw) with Debian and Gentoo is the real first step to have something real.
    I have a funny idea to pick Gentoo Hurd and see if can run some games under software acceleration, like OpenMW could be ok, looking that it lacks of proper 3d drivers for graphic support.

    💬 0🔄 0⭐ 0
  • Oct 17, 2026, 1:29 AM

    @stefano @raster @FossThought More than 20 years ago, I installed Hurd on a Sparcstation 1+, but it was very slow and unstable. I kept waiting for new releases and improvements, but the Sun was liquidated before that could happen.

    💬 0🔄 0⭐ 0
  • Oct 16, 2026, 12:42 PM

    @stefano I find it ironic and cognitively dissonant that they're both embracing 'AI' (stochastic parrots) but also looking for only 'serious' contributions. 🙄 Wow, we're living in some times.

    💬 1🔄 1⭐ 3
  • Oct 16, 2026, 12:47 PM

    @jake4480 I didn't investigate but I'm quite sure they're getting a lot of money from the people investing on AIs.

    💬 1🔄 0⭐ 0
  • Oct 16, 2026, 3:06 PM

    @stefano oh, that has to be it, right? Always monetary incentives. People are so quick to jump on that. There's no amount of money people could offer me to make me use this stuff 😂🤣

    💬 0🔄 0⭐ 0
  • Oct 16, 2026, 12:58 PM

    @stefano A sense of community can be hard to define. It is more something you feel. I'm running Pop!_OS now and other GNU+Linux distributions previously. The world in general has benefited from both the BSD and GPL approach to freedom. Cheers, -Randy

    💬 0🔄 0⭐ 1
  • Oct 16, 2026, 2:29 PM

    @stefano
    Linux has succeeded on Servers & embedded systems etc despite Torvalds. Not because of him. C was already obsolete when he started in early 1990s. A Linux distro relies on GNU which predates Linux. Nor was it or is it the only non-AT&T Bell Labs alternative.

    However MS is worse on the slop front.

    💬 0🔄 0⭐ 1
  • 💬 0🔄 0⭐ 1
  • Oct 16, 2026, 4:56 PM

    @stefano i mean, at least the Linux kernel was developed by a team of professionals in the first place, rather than some kid in his bedroom arrogant enough to argue with the guy who literally wrote the book on operating system design

    *checks notes*

    ...oh

    💬 1🔄 9⭐ 24
  • 💬 0🔄 0⭐ 3
  • 💬 1🔄 0⭐ 1
  • 💬 1🔄 0⭐ 1