Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Linux Sucks 2021 – The End of Linux Is Nigh [video] (lunduke.locals.com)
64 points by WoodenChair on Sept 15, 2021 | hide | past | favorite | 190 comments


Kinda surprised he didn't mention WSL, on the topic of companies trying to enable transitioning away from Linux onto their own operating systems. On that note, if the end is indeed nigh, then replicating Linux compatibility across as many other platforms as possible - beefing up the existing support on FreeBSD and NetBSD, reintroducing it for OpenBSD, getting it going on something like Haiku - seems like an important next step. Linux the kernel might have numbered days, but Linux the ABI could still have a future.

One thing the video misses re: Linux's prognosis is the breadth of hardware/driver support. Linux is, if not the most comprehensive collection of actively maintained device drivers out there, pretty darn close to it. Other platforms implementing Linux's userspace ABI? Big whoop. It's the kernelspace drivers that will entrench Linux for a long while, and it's those drivers that a Linux replacement will need to provide - whether from scratch or through compatibility with Linux's kernelspace APIs - before Linux's death is even possible, let alone probable. Even Fuchsia is highly unlikely to be the Linux replacement Bryan suspects it to be without substantially expanding its hardware support (which is still AFAICT in its infancy; even Haiku supports more hardware, and that's developed with a budget that might as well be a rounding error compared to what Google's able to throw around).


> but Linux the ABI could still have a future.

This is an important point because it conflicts the doom and gloom claim that something like Fuchsia is just a temporary transition technology that will then dump the compatibility.

You can't just dump the compatibility and remain relevant.

If Google deploys Fuchsia on mobile devices, and then morphs it into something that isn't Linux compatible, it just won't be relevant; it's basically a step of retreating into a niche.

Various pieces of middleware won't work on it; I don't think Google wants that.

> It's the kernelspace drivers that will entrench Linux for a long while

You might think, but it is 100% clear? Drivers can be ported to a different system. There are out of tree drivers out there that use #ifdefs to be compatible across kernel versions due to upstream changes. Drivers can be separated into the payload parts, and the Linux glue, and then that glue can be retargetted.

Nobody wants to do that though if they can just use Linux right?

However, the arguments of Linux accumulating code and becoming less and less maintainable could precipitate into situations where alternatives spring up which take what they want or need out of Linux, without all the other crud.

> Kinda surprised he didn't mention WSL

I was thinking the exact same thing when he brought up Fuchsia, but I think he had a point there. Not a lot of us have heard of Fuchsia. WSL is not a threat to anything because it already runs under Windows. It's not poised to take a huge number of devices famously based on the Linux kernel, and take Linux out of the picture. WSL isn't a transition off Linux and could never be; nobody is looking to transition to Windows via WSL. WSL has little room to break Linux compatibility.

Imagine that, say, ten years from now, no Android, Chrome OS or Google TV or whatever device is built on Linux any more. The kernel is Fuchsia or whatever; there is a C library that is the same old Bionic or a derivative. That will be a blow to Linux.


> If Google deploys Fuchsia on mobile devices, and then morphs it into something that isn't Linux compatible, it just won't be relevant; it's basically a step of retreating into a niche.

Android developers don't care about Linux, it isn't part of the official APIs, unless you are an OEM doing your own drivers and AOSP customization.

As long as Fuchsia runs ART (which they are already porting), and the official NDK APIs, Android users will barely notice it is another OS running on their new phone.


Yeah, drivers for mobile devices are already not upstreamed, and are mostly only available in the binary form.


> Nobody wants to do that though if they can just use Linux right?

That's kind of what I'm getting at, yeah. Sure, you could port a bunch of drivers to some new kernel and driver API, but even out-of-tree (let alone in-tree) drivers don't necessarily lend themselves to that, and the comprehensive collection of in-tree drivers makes that a hard sell.

> However, the arguments of Linux accumulating code and becoming less and less maintainable could precipitate into situations where alternatives spring up which take what they want or need out of Linux, without all the other crud.

Definitely seems like a possibility, though (to my knowledge at least) it doesn't seem like there are any projects going in that direction - I suspect because of the relative instability of the driver API (let alone the absolutely-not-stable-whatsoever ABI) and the more intertwined nature of kernelspace code compared to the userspace ABI/API.


> Kinda surprised he didn't mention WSL

- How could you make Linux suck more? - You install it on Windows!


Actually suck less, it has been my option since VMWare made it possible.


Have to agree, Windows sucks less with Linux on it.


I think the commenter's point is the following hierarchy: Linux > WSL > Windows


I know UNIX since my introduction to Xenix in 1993, used several flavours of it for what is now about 30 years, was a Linux Journal subscriber for most of its existence, yet all my laptops with exception of an aging Asus Netbook don't run GNU/Linux bare metal, rather Apple, Microsoft or Google systems, go figure.


No problem you can run Linux on Windows.

No need to run it on Apple...and "Google Systems" run on linux.

You are perfectly fine buddy, go figure ;)

BTW: You forgot your fridge, mine runs Solaris, have to protect that data, but my toaster is a Mainframe (MVS3.8 tk4-) it needs a good job entry system.


Windows might suck less with Linux, but Linux sucks more when run on Windows


Eh, after using Linux mostly as my primary OS and 100% for servers for nearly 30 years I think Linux is only just now in the last 10 years started to come in to its own. The game isn't ending, it has just started! Linux is more than ChromeOS and Android. Everything Internet is Linux, IoT, routers, appliances, cameras, vehicles, the list is endless.

Fuchsia, being Google, will be as evil as Chrome. And it may be popular just like Chrome and Windows but that doesn't mean there isn't something better for users who care about control, security, and privacy.

There is no way in hell Fuchsia or anything else is going to run the Internet like Linux does. Even the alternate OS's like Fuchsia, Haiku, MacOS, etc don't have server versions. The mammoth Microsoft even failed at servers and they do have server versions.

Barring some unforeseen conspiracy, Linux is going to be around for a long, long time.


IoT is not Linux. Most small embedded devices cannot afford a MMU, or the monstrosity of Linux. Baremetal or FreeRTOS/ThreadX or similar are enough. They don't need that much power, architectural overkill and security nightmares. Most of them are even formally verified.


Why is chrome "evil"?


I guess mainly because in about 5 years time Web developers can write ChromeOS developer instead on their skillset.


so chrome is "evil" because web developers will be able to create apps for ChromeOS in addition to web apps? Or you are claiming that web apps will only work on chrome?


I am claiming that Web === ChromeOS, when Google is done with Firefox, while Apple keeps focusing on native apps.


This talk is far too long and completely unengaging. I'm about halfway through and I would have fallen asleep if my frustration at how irrelevant the points made sofar wasn't keeping me awake.

I've nothing against long talks (this one isn't even very long) but it is too long for two reasons:

1. He's continuously saying with 10 words what he could easily say with 3. It's very artificially drawn out.

2. Most of the points (sofar) aren't relevant, which weakens the overall argument: making less points would make the each of them more focused and compelling.

On the (ir)relevance of the points made sofar:

- He spends an enormous amount of intro time talking up his apparent qualifications and how he's the only human on the planet qualified to talk about this. I don't care. Allow your arguments to stand on their own merit.

- Linux is apparently dying because "they" (the illuminati???) are trying to force Torvalds out of the foundation. What.

- Linux is dying because SUSE (a company he used to work for) is being sold. I've been running Linux for over 20 years and SUSE had already fallen out of relevancy when I started. I remember getting a copy of SUSE on a magazine CD, wondering why I hadn't heard of this distro, searching internet reviews and concluding it probably wasn't the future of Linux. It's 2021 and this guy is applying things he learned decades ago in his career as if they've retained relevancy. In his discussion of companies responsible for Linux health, Canonical isn't mentioned, and the general assumption is an ironically "Windows-ish" take that old corporate sponsors are the lifeblood of the Linux community.

- In person events are cancelled. Somehow the proposition here is that the worldwide pandemic is targeting Linux and Linux alone. What.

Linux may be dying out, and the second half of the talk may have some better arguments but sofar this guy has given me a lot of reasons to completely ignore anything that comes out of his mouth.


As long as Linux sucks less than Windows and macOS - I'm a happy Linux user.


macOS UI > Linux UI > Windows UI

but macOS has weird quirks that bumps them down in the CLI space.

Linux CLI > macOS CLI > Windows CLI

Linux CLI and macOS CLI are really close, I think but Linux beats it.


I agree with this. I genuinely think that linux is the best "developer" machine there is, particularly if you are cross-compiling. You have complete control over everything and can really get to the bottom of problems if required. I really like the fact that I can look at the kernel sources if I need to understand how the &!%$ something works. You don't often need to do this, but when you do, you really do. The one thing that perpetually just pees me off about basically all Linux DMs/WMs that I have used a lot is that occasionally they just hang, horribly, even when the system isn't hosed, and even when you are trying to use it. It's hard to understate how jarring this is as a user, and MacOS in particular seems very well optimised to prevent it ever occurring. I don't quite know how, but if you ever see the seconds not updating on a 1 s tick in MacOS, something is wrong. Even with renice'd -20 xfce on linux on a machine with TiB of ram and 96 cores, sometimes the UI just hangs.

That being said, Apple's documentation is nowadays oft-lamented as being outdated -- I remember it being excellent, including the first ever description of OO code I ever read, as a teenager. (Aside: It involved "a faucet". Not being a speaker of en-US, I remember distinctly looking up that word and trying to understand what on earth they meant, until "tap" appeared later on in my brain). I've never really written MacOS apps but I gather you have to "drink the kool aid" at every available opportunity and to some extent this shows.

My experience of Windows over the last 30+ years is one of confusion, culture clash, a lack of understanding, and frustration. I am not qualified to comment on it, but it does seem to be the case that people can spin up GUIs very quickly very routinely, and they tend to keep running for a bloody long time as well, unlike both other options.


> I genuinely think that linux is the best "developer" machine there is

I agree. And apparently 25% of professional developers also [1]

[1] https://fossbytes.com/linux-market-share-among-professional-...


My company decided that we would use windows laptops for development, even though we deploy to linux servers. Wsl has saved my bacon more than once. I wonder how many devs really have a choice.


Making it a minority opinion...


Linux is "just" the kernel, and statements like

XNU UI > Linux UI > Windows NT UI

don't make sense because Linux has many different UIs. I'm pretty sure there is no macOS UI comparable to my preferred tiling window managers Sway and i3, and in that sense which UI is better is a question of personal preference and habit. In fact, I'm pretty sure only one macOS UI is available. Same for Windows: only one Windows UI available per Windows version.


Most people think of Linux as GNU/Linux. But of course you are technically right.


At the peak of it's game (Mojave), MacOS was really competitive for my workflow. One evening I made a simple realization that kinda trashed it all, though: I was spending more time making MacOS like Linux than I spent making Linux like MacOS. By the time Catalina came out (and trashed 32-bit libs along with it) the value proposition was squandered once again. It's funny how transient the usability of a 'stable' OS can be, even on the Linux side of things.

Of course, Windows is still bringing up the rear. WSL is mighty competitive though, and I'm giddy with anticipation of what Windows 11 will do with it.


Android on W11 is something that was much more possible thanks to the work on WSL and WSLg


Actually all of that is possible thanks to Android on WinRT, aka Project Astoria.


I can't stand the macOS UI. My ideal UI is something Windows 7-like, and these days that's closest replicated by Cinnamon which understands it just needs to get out of my way.

I think the big missing thing with Cinnamon is some sensible tracker integration so inside document searching can work.


> I can't stand the macOS UI

.. it’s got worse and worse IMHO


It's barely changed IMHO. Last year I ran through every release since 10.1. The UI has been remarkably consistent through that period.


MacOS UI polish > Linux UI polish.

MacOS UI choice and flexibility << Linux UI choice and flexibility.

(A voice of a MacOS user from 2031: "Choice?")


> macOS UI > Linux UI > Windows UI

for whom ? I'm in a room with mac, windows and linux computers, going away from my linux machine is a chore and fills me with dread


> macOS UI > Linux UI > Windows UI

By "UI", do you mean desktop environment? I'm Debian user with no desktop environment, only a window manager.

Linux WMs > macOS/Windows WMs

Or by "UI", do you mean GUI apps? I have fewer than 10 GUI apps installed, and I don't think GUIs matter much, especially when comparing OSes.


yeah, desktop environment.


Just looking at UI, though, I think the trend actually favors Linux. GNOME and KDE have made pretty huge strides in the past few years in terms of general usability and polish. Meanwhile I feel like every Mac OS and Windows update elicits nothing but complaints.


Windows software library >>> macOS software library >> Linux software library


Form or function what do we value more?


macOS UI is written in memory unsafe language

Linux is written in memory unsafe language

Windows is written in memory unsafe language

Where is my memory-safe desktop OS?


Does your desktop OS leak memory or not? Who cares what language its written in if it accomplishes the same goal?


Windows, because a large part of the userspace nowadays uses .NET, even if calling into COM APIs underneath.

Followed by Android, because Google goes out of their way to only make NDK usable for games, audio processing, and native methods, nothing else.

But I do agree better stacks would be welcomed.


CosmoOS is written in C# https://www.gocosmos.org/


Bryan could balance his perspective just a tad by dredging up example of old technologies that are used in production, even mission-critical roles.

Not everything that is old is used only by tinkering retrocomputing enthusiasts.

E.g. COBOL has had a much longer run than Linux. Not to deny that its glory days are long gone, if it had any, but it's deployed in important roles, not just something in the hands of retro-tinkerers.

There is also the fact that Amiga is specific hardware. DOS is also tied to specific hardware. Newer hardware still has compatibility to make DOS work, but at the cost of using a 64 bit machine with gigs of RAM as if it were a turbocharged version of an 80286 from 1985. Maybe an 80386, if a DPMI DOS extender is used.

For now, the retrocomputing enthusiast who keeps riding on the back of the supposedly dying Linux horse has a good desktop experience with a brand new computer, which is used to its full potential. This will likely be the case going forward, unless hardware fundamentally changes (like away from the Von Neumann model, type of thing).

Linux will be surely dead when it can't make use of the current hardware, when you need to resurrect (or emulate) legacy hardware to be able to play with Linux. (Even in that situation, it will likely be used in production roles, not just retrocomputing hobby situations.)


Bryan Lunduke revisits this talk every year at linux conventions. Usually good observations.

BTW, he loves linux.


I hate long slow videos? Any slides.

He seems sometimes, if I remember previous years, to forget the point is open source, not linux.

If a better Open Source OS appears tomorrow, I'd abandon linux for it without a backward glance.

If it's closed source, it could be rainbows and ponies and I'd still turn it down.

On the otherhand, I bet an Open Source OS could grow any rainbows and ponies I actually need and want.


> to forget the point is open source, not linux.

For you, perhaps. But for most companies, they use Linux because it's the best server OS around in regards to performance, compatibility, customizability, footprint, support, etc.

Sorry BSD folks, I know BSD is great too, but it's just not as widespread as Linux and therefore doesn't quite have the compatibility and support Linux enjoys.


"... but it's just not as widespread as Linux..."

Of course "BSD" means more than just a kernel. Nevermind all those OSX and iOS users. Companies are interested in more than just the BSD kernels. Many years ago, I remember someone from Google approached NetBSD asking about their libc.

Most "BSD folks" seem to know a lot about Linux. Yet I dont get the same feeling from "Linux folks". I have never learned anything about BSD from "Linux folks", but I have learned various things about Linux from "BSD folks".

NB I use both. I do like Linux.


You missed Playstation as well.


Plus the old AirPort Extreme and some Cradlepoint routers. Plenty of other examples Im omitting.


When it's about open source(somehow UNIXy, maybe POSIXy) compatibilty shouldn't matter(much).

Because it one rebuild from source, and it's there.

Ah, yes, forgotten concepts from long before the war

of eternal attrition.


The point is it became that _because_ it is open source.


> The point is it became that _because_ it is open source.

I still disagree. Most companies happily pay for their server OS's and support (Red Hat, SUSE, Oracle, etc), and don't even think about the open source aspects.

Maybe a long time ago it got traction from being open source, but so is all the BSD's and they aren't nearly as prevalent as Linux is. Something else has to be a factor here besides it being open source.

Perhaps you mean it became that way because open source attracted talented developers? Perhaps... although today most contributions to the kernel are from paid developers working for mega corps like Google, Amazon, IBM, etc, who do that as part of their day job.

As an aside, I love and support open source projects, I just don't think it really matters for Linux (at least anymore). Plenty of folks would happily use it if it were closed source and proprietary - it's that good today.


> I still disagree. Most companies happily pay for their server OS's and support (Red Hat, SUSE, Oracle, etc), and don't even think about the open source aspects.

This is only partially correct, IMO. I'd argue that both are happening:

1) Companies are willing to pay

2) People in those companies are thinking about open source aspects

The key is that it's not necessarily the same people. The buyer probably doesn't care about open source, and it's not the sysadmin/developer/etc. that worries (or doesn't worry) about the cost.

Companies pay for open source operating systems because they're running a business and need support when things go wrong.

Sysadmins ask for open source operating systems because they do care about the ecosystem and how open source made Linux what it is today.

There's a reason a former employer of mine that ran a mix of AIX, Red Hat and Windows gradually moved more and more towards standardizing on Linux. The AIX boxes where incredibly difficult to manage because of ancient libraries and lack of access to modern versions of common tools. Lack of tools = sysadmin pain, sysadmin pain = someone pushing leadership to buy a better operating system.

At the end of the day, Linux was just simply better. I think a case can be made that it was better because of the open source ecosystem backing it.


This just tells me that the person "pushing for a better operating system" didn't have the technical chops to compile, link and cleanly package software for AIX she or he needed, so it's a matter of throwing money at the problem and buying an inferior product (GNU/Linux) because of incompetence. Happens very often and predominantly in IT.

SEE ALSO

"A Brief History of Sadness" chapter

http://dtrace.org/blogs/wesolows/2014/04/10/libsunw_ssl-or-h...


Why would you conclude that GNU/Linux is the inferior product here? The year was 2009, and the Linux ecosystem was in a very different place.

Technical chops and "the OS has such a poor set of build tools we need a specialist" are two very different things. If you follow this line of reasoning too far, it quickly descends into absurdity: "Buying an operating system just tells me the person didn't have the technical chops to write their own" is something that I'm fairly certain neither of would use as an argument.

If a product is "superior" but is so difficult to use that people abandon it, was it ever superior to begin with?


I'd argue it became that because it is free (as in beer).

People who do things for ideological reasons are always in the fringe and not representative of the whole.


OpenBSD might not have rainbows and ponies, but it has Puffy and it usually compensates more than enough for lack of those.


Does Lunduke use a teleprompter and read his blog posts at conventions verbatim? I used to subscribe to his YouTube channel, but it eventually just became blog spam that he repeated in video form.


On the plus side, rms returned to the FSF about a month after this video was made. Maybe things are slightly less nigh?

https://www.theverge.com/2021/3/22/22344910/richard-stallman...


I don't understand why that would be relevant, Richard Stallman has nothing to do with Linux and has never worked on Linux at all.


The linux kernel is only one part of a functional system. RMS contributed gcc, gdb, emacs, gmake and more into the gnu system, which are all used in most distributions.

That said, I think the GPL is his most important contribution, because that has kept such projects (such as the linux kernel) available to all.


BSD is also available to all, and it uses the BSD license.

The GPL is an interesting thought experiment but honestly I don't think it makes much difference these days. Apache seems to be the license that won over corporate America, and people are getting a lot of benefit from those projects. (Meanwhile, the cloud happened and no software licenses were ready for "AWS decided to run your thing for money and not pay you for writing the thing".)


I wonder how many BSD enhancements have disappeared into corporate source trees because of "fork and close".

(not that the linux kernel has necessarily benefitted from every corporate tarball downloadable from their website)


And all are replaceable. The kernel is not.


That's the GNU system though, which is a different thing entirely. It has always supported multiple kernels, in fact a lot of GNU software has ports for Microsoft Windows or can otherwise be cajoled into running there.


Saying that Richard Stallman has "nothing to do with Linux" is like saying that Naughty Dog have nothing to do with the PlayStation. They don't work on the hardware or OS, sure, but they do happen to write some of the most popular software on the system.

GNU software installation counts are highly correlated with Linux installation counts. A user's overall experience with their Linux system is somewhat correlated with their feelings about the GNU software on it. If people enjoyed GNU software 50% less, it would certainly harm Linux.


I'm afraid I don't understand your analogy, all of that has even less to do with Richard Stallman, who does not work on GNU software and hasn't done that in decades.

Edit: I had to look up Naughty Dog that made me more confused, they only make Playstation software, but as I said in a sibling post: GNU has always supported multiple kernels.


You're just being intentionally obtuse. The original comment's argument was that Richard Stallman's return might result in higher quality software from the GNU project. That clearly affects Linux, since large amounts of GNU software are almost without exception deployed alongside Linux.

Yes, there are two "hops" in the relationship, but it's still not much of a stretch. It's not hard to believe that Richard Stallman's presence at the FSF will influence the development of GNU software, and it's flat-out obvious that the quality of GNU software impacts Linux, just like the quality of games developed for the PlayStation impact its success.


>You're just being intentionally obtuse.

Please avoid assuming bad faith, this is against the rules here, and it's pretty hostile to me and makes me not want to talk to you.

I don't see how that argument follows, as I have said, Richard Stallman does not develop GNU software, very rarely influences its development regardless of his position, and also one does not need to have any presence at the FSF to influence the development of GNU software. In general the FSF does not drive development of GNU software, the contributors do. And many GNU contributors are not involved with the FSF at all (similar to how many Linux contributors are not involved with the Linux Foundation).


> ...it's pretty hostile to me and makes me not want to talk to you.

Evidently not to a sufficient extent.

That the FSF and GNU are more closely connected to Linux than they are to Windows and macOS is self-evident. Roughly 100% of Linux users are extensive GNU software users. And Linux uses a license whose authorship and legal defense both fall under the purview of the FSF.

Would you argue that Richard Stallman, the founder of both the FSF and GNU and the author of both gcc and emacs, might not still do something to benefit the FSF and GNU software--and therefor all Linux users--in his remaining life?


>Evidently not to a sufficient extent.

Please avoid these snarky comments, this is also rude and unhelpful and against the rules. I'm continuing to talk to you as a show of good faith even though you're discouraging me from doing so, I ask that you please be respectful to me and maintain the same show of good faith. We can disagree and still respect each other.

>Roughly 100% of Linux users are extensive GNU software users.

If you meant just people using the Linux kernel, then this is very wrong. Android is the majority of users and does not really use any GNU software. If you meant people using Linux on a desktop or server, then it's still probably not 100%, there are various non-GNU distros. Also this is really not related to the FSF/GNU relationship at all so I'm not sure why you brought it up.

>And Linux uses a license whose authorship and legal defense both fall under the purview of the FSF.

This is also false, the defense of the license of Linux falls to the authors of Linux. The FSF has no bearing on their actions, they can attempt to influence them but it carries no more weight than any other outside group. For some obvious examples: the FSF asked them to switch to the GPL3 but they declined. People have asked them to follow the FSF's interpretation of the GPL, but they declined. In any case I think you may be confused as to what the FSF actually is.

>Would you argue that Richard Stallman, the founder of both the FSF and GNU and the author of both gcc and emacs, might not still do something to benefit the FSF and GNU software--and therefor all Linux users--in his remaining life?

I'm sorry, I don't understand this question. Many people could theoretically do or not do things at some point in the future to benefit any given piece of software, so any answer would be purely speculative. Also, enhancements to any given GNU software do not benefit all Linux users. There are a large number of GNU packages that aren't popular at all and aren't even available on some Linux distros. Also, Richard Stallman is not the sole author of gcc and emacs, and AFAIK has not contributed code to either of those projects in many years.


> I'm sorry, I don't understand this question. Many people could theoretically do or not do things at some point in the future to benefit any given piece of software, so any answer would be purely speculative.

It is no longer possible to assume good faith. Goodbye!


Yup. The video creator thought it was relevant (~7:00). I think it's a little funny he was back on the board a month after the video.


relevant link that would otherwise have two of these contradictory copypastas: https://wiki.installgentoo.com/index.php/Interjection

edit: the whole point is that people often mean Linux as in some distro and not Linux the Kernel


I don't understand why this is relevant. To use the word from that article, please do not "shitpost" here, let's keep the conversation above ground.


No. He has nothing to do with Linux (as in the kernel). Parent is right to call it just Linux.


Well, Linux definitely sucks.

By a chance - is there anything better? Pretty please?..

Actually in case Fuchsia manages to replace - well, I would be happy. But would it really happen?..


Linux is like Democracy, it’s the worst except for all the alternatives.


You may want to check out the *BSD OSes like FreeBSD, OpenBSD and NetBSD. Just try them. Watch some videos to see what they're about. They might very well appeal to you. I personally think they are great and am embracing them (particularly OpenBSD) over Linux for most new computers I set up.


I want to know the answer to this. I have a really good track record of picking things that are about to get hot (julia, elixir, zig). As much as I love linux, I also hate it. I'm ready for the next OS.


Noone is working on the next general purpose OS for the end user. Maybe apart of Google.

For some reason people wish to stick with dated unsafe crap from prehistoric era.


GNU Hurd should be ready any day now...


They got audio support in a basic form a few years ago. That's how far behind they are.


Ready for what?


For being put out to pasture?

Right, I'll show myself out...


You may do some more research [1].

[1] http://www.barrelfish.org/


Looks like another pile of memory-unsafe C crap: http://git.barrelfish.org/?p=barrelfish;a=blobdiff;f=usr/dri...

This will never become a modern OS. Wasted effort, wasted time, wasted money.


[something].js https://xkcd.com/1508/


M.E.T.A.L (https://www.destroyallsoftware.com/talks/the-birth-and-death...) - hopefully "pandemic" won't take the place of the 5 years war between 2020 and 2025


That depends pretty much on the anti-vaccine folks.


> Actually in case Fuchsia manages to replace - well, I would be happy. But would it really happen?..

Would you be happy? Google has a poor track record of maintenance, not to mention monstrosities like angular, golang, kubernetes... I shudder to think about what a Google OS would look like.


Well, fuchsia seems to be better designed and from what I've heard they had intention to rewrite their kernel in Rust. Eventually.

Fuchsia seems to be FOSS.

So, in case a new userland appears on top of this legacy-free system I, probably, would be happier than I am right now on damn Linux where nothing works as it should and everything segfaults.


Problem is that the software license has become irrelevant for the largest tech companies. In the old days there was a fear among them, and probably rightly so, that their software would be "stolen" if they went open source.

But today every large corporation is running projects as open source without any fear. What happened? For one these projects are so complex it is not feasible for Joe the programmer to read thru all of it, it takes a company of the size of Google or Microsoft to manage it.

The other thing is that the business model changed. Now they give stuff away for "free" to lock you in and spy on your every move and then monetize it.

Third is the close integration of hardware and software, so even if someone forked Fuchsia it would be utterly pointless because of hardware.

What Google has shown is that it is possible to operate one of that largest spying agencies in the world all on open source.


They still don’t open source their secret sauce.


If they would I'd call that business model open sauce


I think rust is a singularly poor choice to write a kernel, due to the datastructures required, poor support and interop between libraries and nostd, in the long run macros obscuring code and making understanding timing attacks difficult, e.g. but that's just my opinion... If you have the resources of google, you might as well write it in another language and apply a proof solver for security assurances (or just use SeL4).


Windows


What's your definition of "better"?


Sorry, but what is "Windows"?


Nice little OS, has been around for a bit. Good backwards compatibility to boot.


I had a look at it, looks like another inconsistent pile of unsafe C crap stirred with tons of legacy frameworks and complete incoherence.

Doesn't look like an OS at all.


Windows is a mix of legacy C code, which started to be migrated into C++ around the Vista time (hence /kernel in VC++) and .NET/COM (WinRT is basically COM with some extras).

Modern kernel code makes use of WIL.

https://github.com/microsoft/wil


> around the Vista time

I remember stumbling on the win 3.11 sources on github at some point, C++ was already used back then


For what would be "userspace" yes, but they were crazy enough to still do COM in C, given that by Windows 95 C++ was still unwelcomed on the kernel.

https://devblogs.microsoft.com/oldnewthing/20210824-00/?p=10...


One that has a bit of an identify crisis regarding it’s UI.


It does, and several teams seem lost in political battles trying to make their own stack the next UI while handwaving the chaos, yet Win32/Forms/MFC/WPF aren't going away, if one wants to code something today that will keep on running.

I bet Windows 11 will still provide the VB 6 runtime.


Why is Windows better than Linux?


"It just works" is a factor for a lot of people and businesses.

Whether or not this is possible because Microsoft is an aggressive, monopolistic company is a different conversation altogether.

I pulled up Azure last weekend to see what the pain level would be on standing up a new corporate environment for our organization. It turned out to be so easy I just did most of it on Friday night. The thing that was most staggering to me was the ease with which I could provision AD DS services. I don't even know where those computers are. I am more than happy to not have that problem on my hands anymore. Some of you will argue that this is a perk of Azure and not Windows, but I would say that Windows is the enabling factor behind a lot of this automation.

On Monday morning, I handed everyone new windows credentials and it all just worked. No fucking bullshit. VPN, certs, office, outlook, visual studio, everything.

For organizations that don't have time for bullshit, Windows seems like a reasonable option to me. I know a lot of you can come up with an unlimited supply of one-off reasons why windows should never be used, but the harsh reality is the entire ecosystem considered together is fucking wonderful if you don't have an unlimited IT budget and need to turn a profit to survive.


I'm in the middle of installing Windows for business purposes, and this is far from my experience. I've yet to see it "just work."

Installation was a hassle! THREE hours just staring at a various setting-up screens. And then it has the gall to bully me for using an offline account; it screeches about my "diminished experience" because I refuse to accede to its native advertising. Want to uninstall Edge? Sorry, your moral superiors at Microsoft decided it's better for your own good that you don't.

Now I'm having to jump through hoops to figure out how to get it to use a shared directory, all because Microsoft decided they _just_ had to cut its own path through the CLI jungle as opposed to just using bog-standard bash.

The ecosystem seems user-hostile and full of more bullshit than any amount of Linux. I just don't see it.


Nit pick: Bash is only standard on GNU, it has never been standard anywhere else. You may be thinking of the Bourne shell, although neither bash nor the Bourne shell include a standard for shared directories, so I'm curious what you mean is bog-standard there.


I'll have to review my shell terminology; however, I'll assume you're correct here, and in that case I'll eat crow on both points. I appreciate the nit-pick! It's good information, and an obvious blind-spot that I should improve upon.

The "bog-standard" I'm referring to is my ability in various Linux distros to easily find and mount network-shared directories. I assumed this was some power of bash, but if not then I'll reflect on that and find whichever utility enables the ease of my workflow.


Are you mounting them on NFS or using samba? That wouldn't be bash. Under windows you would just type the SMB path name.


Well, better is subjective but a couple of reasons that make it a default choice for most use cases:

- User familiarity

- Prevalence

- Pre-installed

As an aside, I’ve installed plenty of Linux instances on people’s old hardware. In situations where I’ve set it up so everything “just works”, and where they were not big Windows users in the first place (so didn’t have to re-train muscle memory/find things) I found they got along just fine.


Man, this does make me wonder if Linux today is the FireFox of 2007.... An absolute juggernaut that is far superior to its cooperate competitors (IE/Safari), but can't quite comprehend what is about to happen when Google releases this little thing called Chrome...

Fuchsia is definitely a vastly underrated project (just from the fact that it has survived 5? years without the overlords at Google killing it off should indicate something).


Fuschia will survive (I assume Google wants to have an operating system with a license that isn't GPL).


Could fuchsia, as a google product/project, get the kind of traction that has contributed to the success of Linux? I think it possible, but somehow it doesn’t seem realistic. Would google’s competition flock to using google’s proprietary Linux replacement?


Fuchsia is already in production, the Nest now use it.


Linux has had an amazing run, considering Linus is on the wrong side of the debate with Tanenbaum about microkernels. It is going to implode.

Genode, at least, and likely the others, have a process for converting linux drivers into their own systems, of course there they aren't in the kernel. Because of this, the costs of getting a new OS up to speed aren't huge.

Fuchsia, Genode, and GNU Hurd are the top contenders in my opinion for the next step in systems. We've been layering band-aids to everything to try to make up for the lack of capability based systems.

Bold prediction: Once we get a capability based OS with a reasonable amount of adoption, all the containerizing of applications can go away, and you just throw things into a directory, and run it with ONLY the resources needed to do the job.

I, for one, won't miss Virus scanners, and the constant worry that a program might trojan horse everything.


Their summary:

- Founders banished

- Community in shambles

- Companies losing Linux focus

- Core system components reaching sustainability

- Google building replacement

- Statistically unlikely to stay active much longer


Hint: The community has always been in shambles. It's a feature, not a bug.

Founders aren't banished, they aren't in any gulag. They can carry on. If they produce something better, they are still the drivers.

A commentator below said, "remind me to make my own license with "I take it all down with me if you destroy me" clause....

That is explicitly _not_ and open source license.

The point is no one, neither the founders, nor any that may replace them, "take it all down with me".

You may get a fork, well then, may the best fork win.

A while back gcc was forked. Remember egcs? It was the best fork for awhile.

Then the fork merged again with gcc again.

Yup. It happens.

Life moves on.


Yeah, this is almost like criticizing democracy for being messy. The mess is sort of the point. :-)


Quite verbose for such limited content.

Anyway, Linux has always been the under-dog. It was "statistically" much less likely to survive in the late-90s and early-00s.

In my personal opinion, Linux offers me the best desktop experience today.

I think I'm not in a bubble, I dabble with MacOS and Windows often. Linux feels liberating. It takes time to customize to one's preferences, but after that, it is smooth AF.


> It takes time to customize to one's preferences

I don't even do that. I find the {mint, PopOS} defaults to be incredibly sane. I think I changed the background of mint. I'm always SO very confused when I need to use a MacOS or Windows machine.

That said, you can get the feeling out of the corner of your usage that something's gonna give. There has been painful transitionary stuff going on (remember X vs wayland? ifupdown vs networkmanager vs netplan? systemd? musl vs glibc?) that somehow the community papered over in a sane fashion for GP use, but when I was "sysadminnnig" stuff as an accessory to the stuff I was coding, it was loads of pain normalizing between different machines and moving from distro to distro. This could get even worse as stuff moves to containerized work, so the people most likely to complain due to bearing the most pain just stop caring.


> Anyway, Linux has always been the under-dog.

It's currently dominant on phones and servers. Google can snap its fingers and set one of those on a course to drop toward zero over a few years, and is well-positioned to push Fuchsia for cloud-related server purposes if it so chooses, so that'll likely go too, if they decide to do that. AFAIK most Linux desktops are ChromeOS devices, and that's Google again.

That leaves traditional Linux desktop users, but if Fuchsia's GUI is any good and is well-integrated, and driver support is OK, those'll bleed to Fuchsia faster than they're replaced, too.


I could see Google wanting to get away from Linux, especially if they build more of their own silicon like Apple has, but I'm not confident that'll translate to servers outside of Google moving away from Linux.

I can definitely see Google wanting to consolidate their codebase and switching all the consumer devices they maintain to Fuchsia sooner or later.

As a part time Linux desktop user since ~2002, I don't think other traditional linux desktop users would bleed to Fuchsia. I could see some people switching from macOS to Fuchsia though.


I've also been using Linux for about 20 years and I would happily switch to Fuchsia if it had good hardware support and worked well on servers. POSIX is just a bad API spec, I'm personally sick of programming for it and dealing with it. It seems to me that new things like eBPF are just papering over the existing problems, which everyone is afraid to address out of fear of breaking legacy software.

But I actually don't think this will happen with Fuchsia and I expect to just deal with the problems of Linux for another 20 years. Oh well.


The thing is Linux underpinning Android and ChromeOS is largely relatively irrelevant for the Linux community. It's convenient now and again, but not many of us rely significantly on either of those for our use of Linux. Replacing those uses with Fuchsia would mean nothing to me, for one.

In any case, given Google's lack of attention span, I suspect Linux will still be going strong another 4-5 new OS's from Google down the road.


the bluetooth can be annoyingly wonky at times for me, but otherwise Linux is definitely the best in my experience


I don't want to put words in Lunduke's mouth, but since it seems to be a politically charged issue, I think that in this video the perspective of why "Founders banished" matters has nothing to do with whether or not you agree with why any given one of them (or all of them) have been banished, the fact that it's all happening at once - whether or not it's a coincidence - suggests a bad future for the ecosystem.


I disagree, out-of-touch founders being kicked out is a good thing. It's a sign that the community is strong enough and can move forward irrespective of any one person's opinion. This video has it totally backwards.

Like, I get that there is an aspect of founder worship on this forum but there is no inherent value in this beyond the idea that someone can keep control indefinitely, which doesn't matter for an open source project. Maybe it matters if you want to hang onto your company's bank accounts but that's a fundamentally different thing compared to a project that exists entirely on public github.


You may have a point in theory, but the harsh reality seems to disagree.

Kicking out stupid people may be all good and dandy, but I'm very wary of the people who are harsh proponents of the kicking, whom they represent, and what they stand for.

Oh, and what/whom they suggest the "stupid" leader X should be replaced with. Usually their thinking doesn't go that far.

The usual pattern I see is: employees at large corporations attacking people who don't have a big corporate affiliation. This may not mean much, but it makes me start thinking about this big picture.

And then, I haven't yet seen an open source project where the long-time leader went away because of burnout of some kind or was kicked out, and it started growing, prospering, and kicking ass in general (Python doesn't count, since Guido never really went away). I've seen stagnation at best, and disuse, lack of maintenance, and a fall into obscurity in the vast majority of cases. In other words, there is not enough laissez-faire in "the community" at large to keep the tragedy of commons at bay.


>employees at large corporations attacking people who don't have a big corporate affiliation

That's interesting because on this forum I have seen quite a lot of the exact opposite: employees at startups attacking big corporation employees, seemingly because they (incorrectly) perceive them as their competitors. In fact any time Facebook, Google, Amazon or Microsoft seems to get brought up here in an article there is a barrage of negative comments and airing of grievances out of the blue. Honestly to me this is all just shades of the same open source drama, it really doesn't matter in the end who stands for what or represents what. Some people just won't get along and that's what the community has to deal with. When it's a prominent founder causing issues, then it becomes a problem for everyone.


I agree that kicking out out-of-touch founders is a good thing.

But let's not conflate a value judgement with a prediction: When it happens all at once it says that there's either something wrong with the zeitgeist or something wrong with the leadership, or something anyways which is a leading indicator of the end.


Or maybe it's just a coincidence? Please be careful not to get drawn into "doomsday" thinking.


Bryan Lunduke makes some really good points especially regarding Google, that Google eventually will replace Linux. It is worrying how Google eats up every market and at the same time branding their actions as good will, "hey it is open source!"

If the US ever wants to keep some of their liberties they need to break up Google.

For Europe, we need to build our own web platforms without the American big tech.


European search engines and social networks do exist, they are just not very competitive even on the European market


This thing with the founders is interesting (remind me to make my own license with "I take it all down with me if you destroy me" clause), but I want to know much more.

But who/what is the attack vector? Is it a coordinated attack and if so, what is the method and who are the backers? Are there other founders included in this trend? Granted, those three are pretty big but ...


I'm not sure it matters... Even if it was completely uncoordinated or by accident, it's not good for the ecosystem.


I would disagree. If it were coordinated, it could possibly be exposed and countered.


If it were a coordinated attack then the chance of successfully countering it with a loosely knit community would have been be slim at best.


I think the loosely knit community in question has already lost.


that's a fair assessment, but it's also already happened, so, even if the outcomes are reversible, it's still a bad "leading indicator", as it were.


It's always either Marxist feminists or agent provocateurs for proprietary software. Or both.


I think I would want a few more data points before signing off on the first one, but perhaps not the second.


>Founders banished

Linus was banished?


Not yet. There were some hot headed debates during the height of #metoo about his communication style including calls for him to change his ways or stand down. Linus choose to change his ways and change the code of conflict to the code of conduct. https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin...


So the founder of Linux wasn't banished. Got it.


Not yet, although the golden question is what will happen to Linux, once all key people are no longer around.

Most likely I will even be gone first, but still like most companies, it won't survive a couple of generation changes at the commanding wheel.


He's always had that super-aggressive, rude style in discussions. Linus is not a fairy, but he is super smart.


This was posted 7 months ago [0] with a fairly sizable discussion (74 comments)

[0]: https://news.ycombinator.com/item?id=26203471


This is a long form video, maybe the title should be adjusted to inform that


I don’t follow Linux kernel developments too closely, but if complexity in the Linux kernel is such a significant factor, then what prevents anyone from relying on a kernel without much of that complexity functionality built in (i.e., custom kernel)?

Are the changes described just too integrated? It’s been a while since it built a custom kernel, but I seem to recall knobs for everything.


I'm aware that Brian has been trolling with such a title over the years, but I really think that it's the epitome of clickbait. Totally a personal opinion, but I absolutely hate such farcical titles.

I'd honestly feel more compelled to watch yet another one of those with a title like "What still sucks about Linux"


Agreed.

Not to mention he takes ages to say not much.

There are some good points buried in here, but it takes him over 5 minutes before he's even on topic.


i was grateful for the 1.75x speed option on the video!


He's not trolling or clickbaiting. The title is referring to a future that we are always teetering on the edge of where Linux fades away. The video is a list of the things that he thinks the Linux community should be looking to fix to avoid that decline.


> He's not trolling

Yes he is.

> or clickbaiting

I believe he is.

> …he thinks the Linux community should…

You might be commenting on the wrong post, because clearly that’s not what the title of the talk says.

While we’re on the topic, I think the problem is trying to make a “community” out of every other software a group of people use. I think embedding Linux in Windows is wrong, clearly there are lots of Linux people who think that’s correct. I’m not wrong and neither are they. We just don’t have to be a “community” just because we use Linux. The survival of Linux depends on a lot more than what a random circle of people who use it are arguing about.


Much like Google's Fuchsia, FreeBSD is an alternative that has compatibility for Linux binaries.


The kernel's still monolithic, and one of the arguments in the video was core system components becoming unsustainably large. I suppose the community woes the video mentions might apply equally to BSD, although I have no personal experience with how those work in that part of the world.

If Google developing Fuchsia and (again, as postulated in the video) perhaps moving focus away from Linux is a concern, it might be worth noting that even if there are corporate backers and sponsors of FreeBSD, I don't think there are companies the size of Google, Intel or even Red Hat contributing massively to its overall development. I could be wrong, but I'm not aware of that going on.

Pretty much all of the arguments given in the video could be true for FreeBSD or other Unix systems as well, at least if they had the breadth of Linux in terms of hardware support, size and visibility of the community, and corporate connections.


That is the beauty of the license, Apple, Sony can take what they feel like having, and only upstream also what they feel like.

There are UNIX like OSes that use microkernels, like QNX or INTEGRITY.


NetBSD kernel has Linux compatibility. (NetBSD had it for years before FreeBSD.) For simple applications it works well. Fortunately I prefer simple applications.


But not an alternative when we need memory safety or just damn 802.11ac wifi.


No, it has not, that's a myth. Nothing works there properly.


It works well enough for Playstation developers.


Google is famous for launching projects, spending billions, and then... just dropping them.

And just because Google does something, doesn't mean anyone else will use it.

Not sure why Google doing something, means certain... anything.


I couldn't take more than a minute or so. I think people who believe they have something interesting to say should just fucking say it, if they want people to listen to them. That was awful.


As an aside, it’s great to see content on Locals and Odysee. We need these alternative platforms to succeed to diversify the technology utilities our modern society depends on.


He missed, on IoT space:

- NuttX

- Zephyr (ironically a Linux Foundation project)

- Azure Sphere OS (It might run Linux kernel, but very custom one)

- Azure RTOS

- FreeRTOS

- Mongoose OS

- Mbed

- Arduino


I guess there's a kind of humor in the video. Yet, for me Linux has seen by far worse days. I can,remember how Microsoft didnt want to work with us, because we used communist software ...

Also Linux is the most common OS on two planets in our solar system :)

A 42 min video with content that could be put in 5 minutes (not well reasoned at all, as missing WSL and other systemic threats, like right wing idiots) is just sad.


I wish there were a TL;DR, a transcript, or something.


To quote his major points

- Founders banished

- Community in shambles

- Companies losing linux focus

- Core system components reaching unsustainability

- Google building a replacement (Fuschia)

- Statistically unlikely to stay active much longer


TL,DR?


Ok


Linux has never been scale-able, starting from the days of the Tanenbaum–Torvalds debate. You can't grow a monolithic kernel indefinitely. It's amusing how it managed to reach the state it is now.


Why do you need to grow a kernel indefinitely? Surely there's a limit on what kernelspace is responsible for.


I don't want to grow a kernel. Linux grows by 2M LOC a year, as reported by the author of the video. That's just what happens.


Most of that is driver code... and a lot of new devices and do-hickeys are created every year. Seems normal, given you can plug just about anything into a linux system and it "just works".


That's the problem. The system is very centralized. Drivers have to be constantly upstreamed to the kernel. Any other way introduces a ton of pain for all involved, since Linux has no binary compatibility guarantees.

Compare it to Windows NT kernel, which only have a few essential drivers in it, but you can go online and download drivers for anything you need, and it mostly works out of the box.


What is the problem, exactly? As a user, it's completely transparent to me.

Besides, not 100% of the drivers are upstreamed... see Nvidia for example. It just means nearly 100% of hardware works out of the box with zero fuss or configuration.

The same really cannot be said about Windows, which sometimes still requires drivers just to see hard disks during the install.


This doesn't really mean much to most people, since 99% of drivers are for obsolete or rarely used hardware.

If you have recent enough devices, experience is much better on Windows.

When I built a new PC about a year ago, installation of Windows was a breeze, and all drivers were automatically installed.

Whereas on Linux side, I've tried latest version of two popular Linux distributions (Ubuntu and another one I cannot remember), and they didn't even have the network adapter drivers. After spending some time I just gave up, because it was not trivial to fix it.


NVidia in particular has been painful to update on my laptop, and it doesn't work well with the rest of the system.


Yes, unfortunately they continue to stick with proprietary, what seems like afterthought, linux drivers that must be manually installed.


TBH I really really hate having to do driver updates on windows, vs linux which JustWorksTM


Linux includes all drivers in-tree (in theory anyway) and there's definitely no limit to the number of drivers you need.


can I introduce you to something called x86?


That's an interesting analogy! I think it's going to be history as well, just in a much longer term. Apple has jumped the ship already.


One should think about archiving and being able to reproduce code consistently before ditching technologies like Apple does.


(i'm not arguing it's good, I agree with the video, just pointing out that low-level "inscalability" can be kludged indefinitely, it seems)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: