Linux Crosses 4% Market Share Worldwide | Hacker News Reader

Linux Crosses 4% Market Share Worldwide

linuxiac.com

190 pointsbenkan179 comments

This is desktop market share! Overall Linux market share is like what nowadays?

EDIT: Depending on how you count, its more around 50% according to Wikipedia.

https://en.m.wikipedia.org/wiki/Usage_share_of_operating_sys...

Interestingly the highest market share is in supercomputing with 100% of the top 500 since 2017. This is followed by the server market. "Big Iron" Linus said and the biggest iron it was.

100% if you exclude Microsoft’s cloud? :-)

Doesn't Azure Cloud also rely on Linux in the majority of its VMs?

"More than 60 percent of customer cores in Azure run Linux workloads"

https://azure.microsoft.com/en-us/products/virtual-machines/...

So the Linux share would actually decrease if you exclude Azure ;)

> So the Linux share would actually decrease if you exclude Azure ;)

I think it wouldn't, because that would imply 40+% of windows market share outside Azure, which I strongly doubt is even close to true.

But your point that even on Azure, Linux has a large majority is still valid!

Depends on if you want to file the servers running Office 365 under Azure, which Microsoft does to juice revenue to make Azure look bigger.

Desktop market share is the only important metric.

How so? I find it very interesting to see how many servers use windows or what software is used in embedded applications

True, but this is probably not what was meant (mobile).

The issue with mobile is that there is way less incentive to switch browsers. iOS is a locked down prison, not making it a fair market.

I don't understand why they even split ChromeOS out. Especially with their VM system, a good number are probably reporting as 'proper Linux.'

I can't speak for worldwide usage, but I've lived in 4 school districts in the last 5 years and all required Chromebooks. I imagine actual usage is higher than reported.

In practical terms, ChromeOS is radically different from every other Linux (desktop) distro. The usecase is very different, and the day to day handling of the system is also very different.

It makes sense to track ChromeOS in it's own category.

Howso? I use Brave. If I install Brave in their Linux system, it appears like a native app. It runs and reports as Linux. It doesn't even know it's on ChromeOS. I imagine Firefox is the same.

What i meant is that ChromeOS is primarily an OS for (young) students/kids. The primary usecase is in an education environment, and that is the angle from with which they compete against Windows and MacOS. Furthermore, ChromeOS is arguably much more of a threat to Windows/Office than traditional desktop Linux is. Hence why it makes sense to track it seperately.

It's not so different than a modern immutable distro with Wayland and Flatpak apps, like Fedora Silverblue.

Sure, some of the daemons are different, like upstart instead of systemd. And out of the box it supports Android apps and doesn't allow shell access outside a container.

But I don't see what's fundamentally different.

It's managed and owned by Google. That's what makes it different. I don't agree, but there's a clear us/them divide happening with Chrome OS right now that's also generational.

The vast, vast majority of ChromeOS users are children/students and teachers. ChromeOS exists so that Google has more avenues of data collection, so that they can sell schools on their ecosystem, and so that kids grow up on Google products (Drive, Docs, Sheets, etc) instead of Windows and Office. ChromeOS is a very legitimate threat to Windows/Office. Traditional desktop Linux isn't.

The whole server metric is meaningless. Do VMs count? Containers? Serverless?

The kernel running within Android is meaningless. It's not the GUI or the runtime. It performs commodity functions. It's like saying that PEX has 50% of all indoor plumbing share. It doesn't provide any distinguishing, features and the user doesn't care about it. It's cheap, that's its value.

I would agree that the average consumer does not really need to concern themselves with what kernel their operating system. For developers though it does mean that you know your code is written for linux then you know it will work on any operating system using the linux kernel. You can even take your compiled library and reuse it between the operating systems. For people who care about the success of the Linux kernel then keeping track of its market share is nice to know.

I prefer to look at it this way: every fifth desktop/laptop runs a Unix OS.

Ever since I've upgraded to Windows 11 I am so extremely frustrated with explorer's performance. The shortest time I have to wait for it to react to a double-click is 5 seconds, sometimes it goes up to 30+ seconds when I've used the machine for a couple of hours. I don't mean the time it takes an application to "boot", but the time for explorer to deal with itself until it is able to launch something. Even a simple thing as showing the context menu and then displaying properties if that was selected. That is on a i9-9880H Laptop with 32GB of RAM and a Samsung 970 PRO.

It's intolerable and I'm preparing to move to Linux. Since there are some tools I first need to write in order to maintain my workflow, mostly mouse-gesture-related stuff, I'm still waiting for the Wayland transition to get settled.

If I could stay on X11 for a decade from now on, I would migrate ASAP, but if the Kubuntu team decides to fully switch to Wayland without using X11 being an option, I'd have a problem.

If you wanted, you honestly could stay on X11 for the next decade. I highly doubt it's going anywhere. Also, no matter what your workflow is, you can write a program to do that. I'd guarantee you could write a small script to automate Wayland to do what you need

Is it technically possible to guarantee program-agnostic mouse gesture support in Wayland? I would sorta expect wlroots to support it, but I don't think that is a safe bet for an arbitrary Wayland compositor.

This is my main issue, I need full access to input devices and the ability to drop, alter and inject new events.

On Windows I fully rely on StrokeIt, it has been a huge improvement for me which I've used for ~15 years.

Linux's easystroke [0] is now deprecated (and always was mediocre) and the modern alternative "mouse-actions" [1] is not good at all.

I've coded a replacement in Rust+C which works for my simple use-case but it targets X11 and it's unclear for me how I can get low level access to input devices events with Wayland. I've also read something about security getting increased in Wayland so that one does no longer have access to input stuff (unclear to me).

Then the nice thing about X11 is the ability to remotely use a X11 session, which apparently isn't possible with Wayland (except for RDP or VNC which isn't what I need).

[0] https://github.com/thjaeger/easystroke/wiki

[1] https://github.com/jersou/mouse-actions

> ... and it's unclear for me how I can get low level access to input devices events with Wayland.

You can't as far as I know, the reason it has taken more than a decade to gain adoption is the protocol design turned out to be unusable for desktops (at least, probably other platforms too). I've been avoiding it because Zoom & friends didn't work when I tested them. My understanding of the de-facto situation is that everything sucked for about a decade from the release, then (special mention to Drew DeVault) people started extending the Wayland protocol to cover the original design mistakes.

Documentation is available [0]. If I were trying to understand the state of play for mouse gestures I'd probably start by researching what is going on the wlroots project, that seems to be where a lot of the useful innovation was coming from. The extensions implemented in the compositor look like they matter a lot.

[0] https://wayland.app/protocols/

Zoom and friends didn't work, because they thought they can grab root window content and run with it. By a chance it did work in X11 due to how PC hardware worked and thus Xorg organized it's framebuffer, but doesn't neither Wayland (for obvious reason), nor when hardware uses it's newer features, like overlays. Even on other platforms, they have to ask, and the user might permit that. They just didn't bother to ask on Linux (it is done via pipewire), and didn't bother to fix their approach.

But that gets back to the basic issue here - the answer to "how should Zoom take screenshots if someone is using a Wayland compositor?" turns out to be "Well, you needed to implement a separate server that re-imagines video as streams, like PulseAudio. Then they have to interface with that. Of course, Wayland doesn't support that (it explicitly makes it hard, in fact), so someone needs to design a protocol extension [0] first to make all this work".

That is a very complex answer. On X, the answer is "grab the relevant window". I can see why Zoom gave the whole mess a pass until the community sort themselves out. I'm doing the same thing; although it looks like (checks) 15 years of patience has been rewarded and Wayland isn't too much of a backslide from X servers.

Unfortunately automating I/O still isn't all that clear, I'd have clicking automated before I move over; it probably is possible but it is just a messy situation to figure out how.

[0] Technically not a Wayland extension, they bypassed the entire system. So now "Wayland compositor" tells us nothing whatsoever about whether the system can take screenshots, but we have to look up compositor-specific support for screencapture. The situation is ... technically better than X, but still silly. They could have avoided years of wasted time by just acknowledging that people want to intercept data as part of the protocol. The situation is a mess and it is silly that we can't have generic Wayland screenshot applications.

The thing is, just grabbing the root window didn't work for 100% either. Outside of the permission problem, there were others: at first, people noticed, that the mouse cursor is missing, so the screen grabbing apps had to handle that as an extra (get the cursor shape, it's position, superimpose it on the from previous step pixmap). Then, if you happened to use Xv, the content was missing too, you got the chroma key used instead. This use case was never handled while Xv was used, the problem was solved when it became obsolete.

The point is, you either implemented a quite complex way to get a screenshot, or just used another implementation as a library. For a screencast, you discover a new here-are-lions land, since you are going to read out VRAM for each frame, with corresponding "speed" and CPU load.

So yes, separate server handles that and more for you. Compositors finally can use overlays (that's how they emulate different scales or resolutions for clients like games, while using native resolution of the display). Pipewire/gstreamer implement a hardware accelerated, zero copy screencasting for you, so you don't have to do it yourself, just call the right portal.

Yes, apps have to change. But they already do change, at least on other platforms. Apple does way more changes in their system, and do you see anyone waiting for several releases until they update? They changed their arch for christ's sake, and except for the long tail, everyone updated within few months. Why it is so difficult on the linux side?

Maybe the reason is not that it is difficult to update, but that the linux users accept crap and apple users don't. Do not accept crap by the proprietary vendors, and they will step up their effort. It's that simple.

> I've been avoiding it because Zoom & friends didn't work when I tested them.

If you are talking about screen sharing Zoom, Meet and Slack work fine on Debian stable running Wayland in KDE and Gnome. Teams probably doesn't work on Wayland, but then Teams working on any other platform on a given day is at best a lottery.

There's no native Teams for Linux anymore, only the browser version. Browsers have no issue working with Wayland/pipewire.

Right about now KDE is moving to Wayland. KDE Plasma 6 released a week ago will use Wayland by default, with X11 being an opt in mode. I think that X11 while it might exist for the next decade, it will at best be a second class citizen, at worst a buggy and unsupported mess

And even if they pull X support, do you have to use Plasma 6, 7 or 8 in the next decade?

Sometimes it's okay to use old software, especially if it works well for you.

Yes things that were broken for me by default are explorer, notepad (using it with files over a network), calendar pop up, and start menu. And don't get me started on the UI of the sound/battery/network widget.

Every day I'm reminded of something I think Jo Armstrong said in a talk, paraphrased "Ah, an update! Maybe they made it better ;)".

To paraphrase Foone

"The year on the Linux desktop won't be because Linux got better"

i think there is something wrong with your machine, probably malware of some sort, or whatever this "gesture" stuff is doing. clean install probably indicated.

i agree, however windows does have a way of locking you in to explorer with no practically viable alternative

You'd feel at home at answers.microsoft.com as one of their expert "Independent Advisors".

No malware. Clean install is not an option.

if you are considering switching to linux, presumably a clean install actually is an option.

There's absolutely no way your computer is functioning properly if you have multi-second delays on a simple double-click in Explorer.

I think you need to ditch the tape storage system then and maybe move to something from this century

Explorer seems to have always been plagued by some original sin around how it's event model works. Blocking the main thread until every app involved can feed back which can often require network lookups. You can see them trying to move to an intent system but there's too much legacy for them to move past.

RHEL 9 should give you a guranteed X11 experience until 2032 at least.

>If I could stay on X11 for a decade from now on

I think they all talk about that move but i make a bet no one will actually do it. Remember Nvidia?

And i think anything archlinux/gentoo will stay with x11 AND Wayland for at least the next 10 years.

> no one will actually do it. Remember Nvidia?

It's surprisingly usable these days. GSync/VRR has a bit of a frame order problem but the desktop feels really zippy.

It's just snagging at this point. They fix a few more bugs, improve nvidia-settings, maybe sprinkle on some HDR and that's a lot of complaints ameliorated, at least until we all move to Yutani, or whatever the next graphics server's called.

> I think they all talk about that move but i make a bet no one will actually do it.

RHEL 10 is going to be Wayland only. You will be able to run X11 apps via Xwayland, but not full-blown Xorg server and X11-only session.

Somehow my windows taskbar crashes... And alt tabbing doesn't work properly, I have to manually click on the windows in the overlay. My taskbar is always on top of windows, and they don't respect it's space so they just go under it and hide important stuff. Except firefox, which does respect the taskbar's space on a non-primary monitor, obviously leaving the space, but on the main screen it overlaps the taskbar entirely.

> If I could stay on X11 for a decade from now on

why is that?

I am asking, because so far I have seen old linux users refusing to move because they require some specific x11 feature that is not (yet) supported on wayland. I thought that new linux user wouldn't have such requirements.

Personally for me such migration-blocking X11 feature is "not crashing".

A way this used to be possible, also before win11, was misbehaving custom extensions to the explorer shell. IE, some applications install extra menus and handlers as part of the windows shell. Particularly, some such extensions may fight with each other - you can have one installed, or the other, but not both. I use the tortoiseGIT extension to integrate git into explorer, and a lot of the bugs it has seen over the years, have really been work-arounds to survive OTHER broken explorer extensions. My point being, the problem may come from one of your staple applications, not from the fresh window install(?).

I have several tools installed, also such which hook into explorer, but not more than I had with Windows 7, which I replaced last year with Windows 11 on a much more modern machine.

Two weeks ago I had to boot into the Windows 7 on the old machine (it now boots into Linux by default and I use it occasionally) and almost tears came to my eyes when I saw how fast that 13 year old machine is.

Windows 7 was pretty good as an OS.

What exactly is your problem with Wayland? It works fine.

Two scenarios on my part.

Recently we wanted to play Minecraft. I had to install fresh windows.

I found ISO laying on nas. Installed it. Run update, after fifteen minuts and two reboots update told me everything is up to date.

We tried to run several programs, but no luck. Some dll error with missing entrypoints. Installed several cpp runtimes, dotnet frameworks, still no luck.

The we installed Minecraft, but from a page since app shop was not working. Minecraft said we are missing critical updates and it fetched some.

Then windows update was able to find new updates, we had to go through more reboots, and twice i had to select correct privacy settings, because Microsoft likes my data so much.

It took days to correctly install windows. Could have been easier if I used up to date image.

Second scenario is also about minecraft. On windows it wanted me to create window account for my kids. Required kid account to be working, and there were some issues that it said that windows pin does not work, we could not change it on the machine. I think we managed to change it online by some Xbox online, or other Microsoft page. These pages are just a maze of accounts, setups, game passes.

In the end it was easier just to run it on linux, and put just username and password. No advanced admin account management required.

>Some dll error with missing entrypoints.

Try "sfc /scannow" next time in cmd (as admin)

But overall...trow that image on your nas away.

Yeah I did that and it did not solve anything.

I think the conclusion I made is that this windows gets more in my way recently than Linux.

Newer image could solve some issues, but not all.

>Newer image could solve some issues, but not all.

I am happy that you found a less problematic os...sorry kernel then. And who need's the bedrock edition anyway ;)

Oh and btw have a look at minetest:

https://www.minetest.net/

I mean if you start with Ubuntu LTS 18.04 and try to update it to 22.04, you will be in for a ride, especially on real hardware.

Well I also thought about it. However there is a difference. When I download Linux i know if it is outdated. I see date, i see version.

With windows, when I entered download site I did not know if it is missing anything, or not.

I installed windows 10 ISO, how should I know how outdated it is.

As Linux minut user I had no update problems whatever, even going from older releases. It did not take me 2 days.

You can easily check Windows version by looking at the properties of setup.exe. The command prompt you can invoke during installation also shows it. And I bet it shows up during setup as well.

I can only see this trend continuing in the coming years.

I know people love to get cynical about Linux desktops being niche, but the reality is that desktops in general are becoming niche. Really consider the audience here, in broad strokes I'd divide it into three big groups: school, corporate and public.

The public market for desktops is dying, or maybe it's already dead? The average person, especially the ones on the younger side, will just use their phone or tablet for all their general computing needs. If, for whatever reason, they do need to use a desktop, it will be for work or school, and most of their time on that desktop will be spent inside a web browser. For most people's personal life their desktop operating system just doesn't matter. Video games are an exception here, but will they remain that way?

Schools are broadly switching to Chromebooks, these are technically Linux machines, but really they are just terminals to a web browser. The underlying OS exists purely to prevent students from do anything else with them. Even the cases where schools stick to Windows, this reality doesn't change, the platform of schooling is web browsers now.

I'll admit, none of this is particularly new, but I naively assumed that the legendary stubbornness of corporate IT would be what keeps Windows dominant indefinitely, and that's quite a big audience right? Yet despite working at a big boring Fortune 500 company in an industry uniquely entrenched with Windows, they are now officially offering Linux laptops to developers who want them. Apparently Lenovo officially supporting Ubuntu was a big deal, and since all of our development targets embedded & cloud systems anyway, it was kind of a no-brainer decision for management. We still need Outlook, Teams, Office, and such, but we can do so via Office 365, so there isn't much holding back the transition...

Naturally many niches will remain, I'm not saying Windows will go away here, nor am I saying a web dominated world necessarily equals a world dominated by Linux desktops. But rather than Linux Desktops becoming a niche within a niche, I think that Windows will join Linux, and they both will become equally niche together :)

I would argue, that the dividing line lies elsewhere: between creation and consumption.

The average person does not create much, so for consumption, mobile devices are fine (though it is amazing, what a teenager can do to the photos with just a phone). For any creation, be it media, engineering, science, or even just plain old bureaucracy, you still need either desktops, or workstation, and it is going to stay that way for a while. But the market will be a lot smaller than in 90's or 2000's, since all just-consumers moved elsewhere.

A/K/A "generative computing", which has been under attack for a long time. I believe Jonathan Zittrain's written specifically on the topic, though the closest match I'm currently finding is about the generative Internet:

<https://papers.ssrn.com/sol3/papers.cfm?abstract_id=847124>

Also "The Future of the Internet and How to Stop It" (2008): <https://futureoftheinternet.org/>

Response: <https://cyberlaw.stanford.edu/blog/2010/09/do-we-need-new-ge...>

Cory Doctorow on the war against general purpose computing: <https://boingboing.net/2012/01/10/lockdown.html> (2012)

I'm not going anywhere because I still shudder at the idea of accessing the internet with such a tiny screen.

Looking at the data if you include laptops that doesn't really seem to be the case - more holding steady / rising a bit https://www.gartner.com/smarterwithgartner/pc-sales-bottomin...

https://canalys.com/newsroom/global-pc-market-Q4-2021

Looking around the Pret where I am I see eight people using laptops and two on phones.

If by desktop you mean something stuck on a desk then yeah everyone has laptops.

When I worked at a F500 a decade or so ago, I asked about a Linux laptop and was basically told they would like to support them and were working on it, but couldn't because vendors we used didn't support Linux.

I wonder if that's why support for Linux laptops is finally landing. All the endpoint protection and license auditing and what not finally supports Linux.

4% starts getting it into the threshold of Nassim Taleb's "intolerant minority" to take effect. It is too big of a market share to be served only by a niche player like System76 or Starlite, so we should start seeing all major manufacturers (beyond Dell and their XPS line) to add Linux-only hardware.

And from there, what's stopping them from joining Valve to make Proton work for all the Windows-specific applications that are still used in the enterprise?

> And from there, what's stopping them from joining Valve to make Proton work for all the Windows-specific applications that are still used in the enterprise?

If they tried, which I don't expect, I'd guess adoption would stop it. Microsoft is entrenched with the public sectors and companies all over the world. The cost of switching would exceed any potential savings for years...

There are already directives in some EU countries to choose open source solutions over proprietary ones whenever they are equivalent.

There are already plenty of schools and universities that are using cloud based solutions and just run ChromeOS.

All I'm saying is that we are closer to the point where the customer will be able to treat the OS as a commodity. That by itself works already as a bargaining chip against Microsoft and in favor of customers when it's time to negotiate their contracts.

The switch won't happen overnight, but the trend has been long in favor of Linux. This 4-5% mark is significant and can lead to acceleration.

With current windows shenanigans and data slurping at least there is some sort discussion about it and switching.

In my opinion main blocker which introduces huge switching cost (and even bigger policy problems) is missing group policy alternative.

Lenovo also ships Linux laptops, although their Linux team needs the ability to veto hardware choices that were made in support of Windows. (I wonder if the XPS team can choose their hardware; my developer edition XPS was rock solid.)

Lenovo's Linux team seems to use up a lot of their QA/support resources because they are handed a dozen new laptop designs each year with (for instance) a MIPI IPU6 webcam having no kernel driver, S0ix suspend broken at the EC-level, new types of forcepads and trackpoints and touchscreens and styli and soldered on Wi-Fi card with a track record of crashing in Linux... and told to figure it out.

Plus, I think they occasionally get sidetracked by stupid bullshit like getting the boot logo to not flicker moving from BIOS to Plymouth.

Also, they maybe should have pushed really hard on AMD (and possibly Qualcomm and their BIOS supplier) for having _multiple_ kernel releases that reached the stable "updates" on non-Rawhide Fedora that then broke suspend on several ThinkPad models. Bisect. Blame. Revert. If Linus used a newer AMD ThinkPad, some choice words would have landed on the mailing list.

Considering that safari with my ad blockers blocked access to statscounter, I would be very suspicious that this percentage is much higher. The intersection between linux users and users who worry about privacy and know how to block trackers is big enough.

Good point, Linux users tend to be more privacy conscious.

I would probably not be counted by statcounter either.

Having used Linux for 25 years I really don't care about stats. The community remains healthy and there will always be a demographic who want to use an OS that provides more control.

This article does not mention the release of the SteamDeck at all. I wonder if this could have had an impackt on Linux usage large enough to see in trends like these or if it's just a drop in the bucket.

This is desktop only.

Linux is market leader once you bring in mobile, embedded and so on.

> This is desktop only

So no laptops? Isn’t Steam Deck really just a generic PC with a different form factor? Could (and should they even) StatCounter easily filter out the data coming from it

Do you put it on a desk? At least I don't...

But it DOES have a "desktop" mode...

They're looking at browser traffic. I assume people don't do a lot of web browsing with Steam Decks (maybe I'm wrong, never used one), so they're probably not being counted.

Some data I found on the internet suggests Apple ships about 6 million Macs per quarter, and they're at around 15% market share in the stats linked here. Other data I found on the internet leads me to guess Valve has sold around 3 million Steam Decks in the 2 years since they launched.

So my guess is that if you factored Steam Decks into this they'd account for about another 1% - very roughly.

I've definitely browsed the web on my Steam Deck when I wanted to install a fan patch to a game, but I imagine I am in a small minority. Pretty sure the Steam Deck doesn't even ship with a browser, since I remember manually installing Firefox. And booting into the desktop mode has its own friction, and it's not exactly a smooth experience to use a Linux DE on the touchpads / touchscreen. These days I actually just opt to download everything on my actual desktop computer and just SSH them to the Steam Deck.

I believe that laptops are counted as desktops for this kind of stats. The split is by UI. Mobile is phone and tablets (same UI), laptops and desktops are desktop (same UI).

Servers are a case apart because especially for Windows they have the same UI as desktops even if Microsoft has a no GUI option now. I believe that nearly all Linux servers are command line only.

I suspect that the Steam Deck is considered a desktop rather than a mobile device. Same as laptops.

steam deck would be included in that.

According to Steam hardware survey https://store.steampowered.com/hwsurvey/?platform=linux 35% of Steam Linux users have "AMD AMD Custom GPU 0405" (the custom amd chip used by steamdeck). ~1 year ago it was 21%. That gives an upper bound of 35% from those 4% being Steamdeck. Although actual fraction is probably lower because:

* not every Linux user plays games or uses Steam, but probably almost all Steamdeck users use Steam.

* StatCounter numbers are based on website views. You can browse web on steamdeck but it's not very convenient, so I guess that, unless it's their only computing device and they have connected external monitor and mouse, most steamdeck users do most of their web browsing using their phone or computer instead of steamdeck.

In 1997-ish I failed to install Slackware from 16 or so floppies repeatedly. In 2000 I had a system up and running but could not get connectivity. In 2005 almost everything "just worked" out of the box. In 2013 everything just worked.

I have recently started reflecting on how easy it is to do things now for free compared to the proprietary landscape we had before. Virtually every proprietary software has a (or multiple) free open source alternatives. The potential for human intellectual productivity has never been this high.

Interesting that they are counting chrome os separately since it is Linux based. That's another 2%. So 6% of the desktop market is running some form of Linux. There's another 6% unknown, which I bet might include some Linux users that are a bit savvy about configuring their browsers to not leak information. The Unkown line seems to climb faster than the linux line in the graph; which is interesting as well.

Unknowns are BSDs.

For desktop?

Yes, why? Nearly all my Desktops and Laptops have exclusively FreeBSD on it.

In fact I would argue that FreeBSD is better as a desktop than server (where you may want package stability). FreeBSD update policies such as choice between quarterly (much like Ubuntu, but twice as often) and rolling puts its into nice balanced spot between Arch-like and Ubuntu-like system.

The problem is however poor hardware support. They still do not support Alder Lake iGPU, you'll need to hack the kernel source to make it working. But once up and running it gives you nicer experience than many distros. Fells like a more stable sid.

>where you may want package stability

Oh not you again, i already explained to you that FreeBSD supports old versions ~5 year's of important software like postgresql/php etc. for example postgres 12-16, redis 6.2-7, python 2.7/3-3.9, openjdk 8-22.

BTW many rolling and stop-rolling Linux distros do the same. I don't understand why you try to bullshit others.

I think with your fanboi attitude you should not participate in public conversations. Your style of communication is immature and does not belong to places like HN.

Now having said that, for someone like you, who is unable to understand the reason for LTS distros, like Debian and Redhat derivatives, I am explaining again: if were actually running FreeBSD machines (like I do on some desktops), you would notice it constantly receives small updates, like (random examnple) dav1d codec got updated recently, although I do not really need such an update much, which is absolutely unacceptable in many server usage scenarios: you want to have stable, stale software with known bugs, you know very well and learned to love; the only thing you want is security updates.

Many, many people in FreeBSD community seem to fail to understand that~5 years of support of "old" postgres is not the same as what is done in say debian; you do not want even minor of minorest changes of the version of the package, not even bugfix unless it is a critical or security bugfix.

It is how it is in most server scenarios, deal with it.

>It is how it is in most server scenarios, deal with it.

In your bubble.

>many people in FreeBSD community seem to fail to understand that~5 years of support of "old" postgres is not the same as what is done in say debian;

Ah yeah the debian-magic of patching, please can you tell me the difference? ...actually no, please don't.

>Your style of communication is immature and does not belong to places like HN.

Repeating lies do not belong to places like HN i would say.

>I think with your fanboi attitude you should not participate in public conversations.

I love when you contradict yourself -> Your style of communication is immature

What kind of "bubble" are you talking about? Vast majority of servers run either Redhat derivative or Debian derivative and both boast long term support for even old systems. Ubuntu 18 afaik still receive security updates, so does Debian 10. So there is an apparent and understandable need for LTS distros.

This is how it is today. As I said, deal with it.

Debian has package freeze half a year before cutting a new stable release. A stable release is thoroughly tested and guaranteed to have no breaking issues. "The magic of patching" is there to keep this guarantee for 2 years. Often new packages can have disastrous bugs (such a recent bug in inkscape) and you do not want that on your daily driver and willing to sacrifice freshness for stability and lack of surprises, both good or bad. If you really need fresh one or two packages you can either compile it yourself, install from backports or install a flatpack.

My free course into LTS systems is complete. There will be no further replies in this thread.

I see the server expert talks about inkscape and dav1d updates ;)

BSD is counted separately at a minuscule 0.1%. I imagine the proportion of that 6% unknown might be similar.

I'd imagine it's only as much Linux as Android and FireOS are, which might be argued aren't desktop OSes. But for that matter, ChromeOS might in turn be arguably closer to Android than what would be considered desktop Linux.. (disc: have not used ChromeOS)

It runs a linux kernel, which makes it a Linux based OS. It probably integrates a lot of linux stuff too, and it of course comes with virtualization support (leveraging the support for this in linux) to run proper linux applications and development tools, etc.

It's big enough to be worth counting separately though. Interesting that regular linux now has twice the market share of Chrome OS.

It still isn't GNU/Linux, and if ChromeOS virtualization counts as Linux, so does WSL, as the technical approach to run GNU/Linux inside its own VM is quite similar in both OSes.

Normal ChromeOS users only have a browser based userspace.

Semantics. GNU/Linux is a subset of linuxes. Chrome OS runs a linux kernel, includes lots of libraries and other things from linux, which, like it or not, makes it another (and pretty significant) subset of linux. Which is why it is counted separately.

Semantics that matter, unless we are counting uses of Linux kernel.

Like it or not, there is nothing from GNU/Linux that normies get exposure to when using their beloved Web Apps on ChromeOS.

Maybe we should include JSLinux into that, to bump numbers up.

If it gives us another cent, then what the heck :>)

Do you think normies are using GNU software on Linux desktops? Or do you think they use software like Dolphin, Nautilus, Steam, Chrome, Firefox, Plasma, etc. A desktop environment doesn't expose GNU/Linux. It requires people to open a terminal to get to it. ChromeOS has its own terminal that uses bash and has its own set of coreutils for people to use.

If they are using GNOME stuff they definitly are.

thing is, on chrome os you can enable debian app support(in two different ways), you basically get both linux apps and android apps, on android it's not that straightforward. Idk how is this done under the hood, but from an user POV chromeos is closer to a windows with wsl2 support with friendlier experience

ChromeOS started as a very weird thing, but increasingly they're moving toward being a reasonably normal GNU/Linux as we know it, with Wayland. And of course a good Android environment.

I use ChromeOS (Flex on a Lenovo). I believe not counting it as "proper" Linux is more ideological than technical.

Regarding WSL, I have used that too, and happily so. But it's more understandable to count it as not-linux, because of many factors including the Kernel and the file system.

It's probably obvious to most, but I wish headlines like these would clarify _desktop_ market share. Linux has had majority market share in the server space for decades now.

and mobile

Android's OS isn't Linux. In same way my OS isn't Java. It runs Java, but that doesn't make Java an OS.

Android is based on a modified Linux kernel

There is not much of a point to pointing that out considering popular server Linux OSs like Debian and RHEL also use a modified Linux kernel

But you would count Debian and RHEL as a Linux, so is Android.

Debian and RHEL have classic GNU userland (thus, GNU/Linux), Android does not. With normal Linux, it has only common kernel (even though heavily modified) and maybe musl (as an alternative libc).

It's also based on modified JVM. It still doesn't make Dalvik my operating system.

Android OS can best be described as a ball of mud that evolved soup of (semi?-)proprietary firmware.

It was never JVM. Converting to dex happens on developer side, and that severs anything JVM-like.

Nitpick that doesn't affect my argument. Argument is based on access to underlying hardware not it being JVM.

Since edit window has passed:

Just because it's based on JVM compatible VM, doesn't make ART/Dalvik my Operating system either.

Because neither JVM nor Dalvik are OS.

Android is a kind of Linux just like Dalvik is a kind of JVM.

Ok, but what makes OS an operating system? Versus a virtualization mechanism or an app?

In the case of JVM threading, file handling, memory management etc. is delegated down to the host real operating system.

Same happens for Linux, power, memory and bunch of other stuff is delegated to the SoC firmware.

By that standard no OS is Linux. Linux when used in reference to an operating system typically refers to the family of operating systems that use the Linux kernel.

I'm glad you can read my mind and decide what I'm using for OS definition.

But no, Debian Linux running on a VM is not its OS, but Mint Linux running on my hardware is the OS.

An OS is a body of software that:

- multiplexes machine hardware resources

- abstracts the hardware platform

- protects software principals from each other

Does Debian Linux in a VM multiplexes the hardware resources? Not really. Does a VM running on Mint Linux? Again not really. They are multiplexing the resources given to them, but not actual hardware resources.

Android runs Linux on the hardware of the device.

Android 13 did add the ability to run Linux in a VM, but the host OS still is running Linux without virtualization

Linux on Android doesn't multiplex or protect software principals from each other. It did for a brief time, but that led to massive vulnerablilities. Because Linux view of hardware (nice symmetric system with homogeneous cores running single OS with shared memory address space) was different from what was the actual state.

For that reason hardware was abstracted away from Linux and it's only used to run Linux programs. It's not containerized. It's isolated to an enclave where damage it can do is minimal.

Android OS is essentially a ball of propietary firmware mud. With Linux programs and Dalvik JVM.

For more: https://youtu.be/36myc8wQhLo

>Linux on Android doesn't multiplex or protect software principals from each other.

What? Linux is the only part of the system that directly talks to the hardware and multiplexes it between other applications on top. It enforces security between different apps on the system from interfering with each other.

>Because Linux view of hardware (nice symmetric system with homogeneous cores running single OS with shared memory address space) was different from what was the actual state.

Linux does not assume homogeneous cores.

>For that reason hardware was abstracted away from Linux

It's not. Drivers for interfacing with the hardware have to be built into Linux / a Linux kernel module.

>and it's only used to run Linux programs.

All running processes on the system are running on Linux.

>It's not containerized.

It uses the Linux kernel to enforce security boundaries which is what containers do.

>Android OS is essentially a ball of propietary firmware mud. With Linux programs and Dalvik JVM.

You can run Android without any proprietary firmware. Dalvik is not a JVM.

> What? Linux is the only part of the system that directly talks to the hardware and multiplexes it between other applications on top.

Not really. Linux talks to the software written for SoC and it pretends Linux is in control. Again look at the linked talk.

Does Linux control power? No. Bluetooth managment? SoC. USB? Nah. It's done by a chip in SoC.

At that point what is the difference between Linux and JVM?

> All running processes on the system are running on Linux.

And all JVM processes run on JVM. It still won't make it an OS.

There are firmware blobs running god knows what on actual hardware. Linux does not control these processes. But it assumes it does. And those processes abstract away hardware resources, pretending to be dumb peripherals, while running their own processes regardless of Linux.

> Linux does not assume homogeneous cores.

Granted this might have changed between 2021 and now, but it's view of hardware probably assumes one central chip with address space and Linux as only OS in control. Something like a PC.

Design of mobile phones is about as different as you can get from that model.

> You can run Android without any proprietary firmware. Dalvik is not a JVM.

Perhaps I should rephrase it. Android's OS is the firmware on the device, not Linux. Because Linux isn't doing the job of the OS. The firmware is.

Dalvik might not exactly be JVM but it behaved a lot like one. It's also not hugely relevant to my point.

>Not really. Linux talks to the software written for SoC and it pretends Linux is in control.

Which is how any other Linux distribution works. I'm not sure what your point is here. Hardware exposes an API for the kernel to use it with and that's how it's always worked.

>Android's OS is the firmware on the device

Okay, and no computer's OS is Linux by this standard. This is just a pedantic argument.

Not really. I mean the point is that SoC and modern mobile/server architecture abstract away hardware which makes it so the OS is slowly getting enclaved and isolated from the hardware.

It didn't use to be this way, and pretending everything is the same is doing no one a service.

4% is probably about right

I'm am a retired old bloke who has used linux for about 20 years since the early releases of ubuntu and linux mint.

I do my best to advise and promote linux to all my elderly friends and have even installed linux as a replacement to windows for a few of them. Once they get used to Linux they are forever smitten. The increased speed and ease of use is nothing like windows complexity.

Even I struggle with windows, that damn search box, what is that really about but to confuse. Windows is supposed to be a personal computer not a time travelling machine, so whats the issue with updates taking a millennium. I got better things to do than wait for that unnecessary waste of time.

It must be brainwashing and propaganda that convinces people that a windows machines are better in some way.

I watch my friends struggling with their windows data collecting machines and I cry inside.

My old life is so much easier, restful, stress free and I have more pension to spend on myself.

Free as in Freedom, too cool!

It would be interesting to know what the market share is among the HN crowd though... any insights, @dang ?

No wonder with the recent Windows UX degradations.

Also privacy degradations.

And bugs.

And performance issues.

Still, office365/onedrive/teams is hard to beat on price for a small business.

TCO of google workspace is substantially cheaper if you factor third party SSO costs.

“sign in with Google” is almost always a lower price tier of major SaaS applications.

I installed a Linux distro on a spare SSD that I had this week, to once more try out doing more development in it and daily drive it for a bit.

I went for Linux Mint because it has the convenience of Ubuntu (and Debian), though doesn't make me use snaps, in addition to their Cinnamon desktop just being all around a comfortable choice.

So far, all of the tooling just works, the desktop experience (even including audio) seems bug free, though I've had issues in the past with my netbook instead of the desktop. Regardless, it uses less RAM than Windows, feels more snappy and for development, I'm very happy with it. All of my IDEs and toolchains work, productivity and content creation software doesn't have issues either.

Even things like gaming seem more and more viable thanks to Steam and Proton, although it's not quite where Windows is now and won't be there for a while.

Either way, I'm pretty glad with how the OS and its distros/desktop environments are progressing.

Congrats for trying Linux again, I'm sure it will end up paying out !

My 2 cents: the best DE when coming from Windows is KDE. Kubuntu (ubuntu + KDE) is joy. It has a windows 7 feel, and is very customizable and stable.

The last time I tried letting Windows rest altogether for a week left me reasonably satisfied as well, actually: https://blog.kronis.dev/articles/a-week-of-linux-instead-of-...

With Windows 11 looming on the horizon as an inevitability, maybe I'll just need to rip the band-aid off and jump to Linux as my daily driver, since my containers and servers both use it already.

KDE has a pretty good reputation from what I've heard, though honestly my previous choice was XFCE because it runs on pretty much everything (my old 4 GB netbook included).

Dupe:

https://news.ycombinator.com/item?id=38853877

https://news.ycombinator.com/item?id=39252801

https://news.ycombinator.com/item?id=39576200

Both traditional desktop and laptop markets have been declining or flat (respectively), and both are anticipated to decline in near future years, if not already. See e.g., <https://convergetechmedia.com/laptop-sales-vs-desktop-pc-sal...> for overall markets and <https://www.cnn.com/2024/02/29/business/dell-earnings-ai-ser...> concerning Dell's recent performance.

To the extent that Linux has always represented an advanced technical subset of the market, the desktop market is effectively shifting further to technical users. "Casuals" increasingly rely on mobile devices --- smartphones and tablets, possibly netbooks. Technical users who are already more likely to choose Linux (or similar alternatives such as the *BSDs) are increasingly concentrated in the desktop market.

I'll note that I'm not one to lump Android in with Linux, as the overwhelmingly prominent Android UX does not include Linux applications or environments, even if those are in theory available (e.g., through Termux).

Whether or not overall Linux usage is increasing, as in total interactive users and time-in-OS (that is, for those using multiple environments, including virtualised environments, what is the split of time between OSes), and the value associated with usage (especially in professional / commercial contexts) isn't clear from the Linuxiac article. I suspect that there's some absolute growth as computer markets grow modestly overall, but that it's less significant than the percentage of desktop top-line figures would suggest.

(Written as a Linux user for well over a quarter century, and an Android user for a decade and a half, along with numerous other platforms.)

I wonder, is it truly Linux desktop, or embedded device reported as Linux?

The percentage is much higher on embedded devices.

Here seriously looking for alternative to Win 11 for small family company becouse buying old computers with Win 10 Pro gives just few years. Best thing I found is invoicing app written in Java LOL. Need to call that company and ask do they support non Oracle Java :>

But other then that, Linux is good, Emacs can generate simple pdf's to print, Perl scripts still able to sum few digits, even git via ssh works like NAS :)

Just that mandated by law things are impossible.

No monopoly at all...

I don't really trust the datasource TBH, according to https://gs.statcounter.com/os-market-share/desktop/worldwide the macOS marketshare went from 21.01% in November 2023 to 15.42% in February 2024. That doesn't sound realistic at all.

You can't know whether to trust it until you know why it happened.

I think these numbers are quite strong. I consider myself a Linux user as that is my preferred OS, but my browser usage is from different OSs: Phone, my work laptop which is Windows and from my Macbook Air.

I think less than 25% of my browsing is from Linux. I would think many could be in a similar situation, so in a way it is possible that the numbers "should" be higher.

Most of the confusion would be avoided by simply changing the article to: "GNU/Linux Crosses 4% Market Share Worldwide"

It has never been easier to install and use Ubuntu on a computer instead of Windows.

So why aren't more people doing it?

Because of SNAPs. Linux Mint is better as it doesn't have SNAPs.

Let's look at this StatsCounter data:

- In one month, macOS share drops from 21% to 16%

- In one month, Linux share goes up 50%

- In the same month, ChromeOS share drops 50%

None of these things are true. This data is garbage. In the last year, "unknown" peaked at 13%, to boot. Ridiculous.

Once Linux can run all my favorite games and future games I will uninstall windows forever.

This has already happened to me, I only play Dota2 and Civilization. And some cool games on PS5 and Switch. I still have Windows, but everytime I boot it, it takes 3 hours of updates before its usable again.

Right when I'm thinking about switching from Linux to Mac. Not for the software but for the hardware.

I just can't find any laptop that can even get near the quality of a MacBook (excellent touchpad, sound, battery life).

2024, the year of Linux.

More discussion a few days ago: https://news.ycombinator.com/item?id=39576200

This is 4% of Desktop, Linux is most of installed systems if you include routers, servers and devices. I have used Linux Desktop for decades.

Essentially Windows sucks. I’ve been using MacOS, but I really wish I could get a generous amount of RAM and HD at a reasonable price.

Steam deck must have booster the numbers a lot

it’d be interesting to know the ratio of consumer desktops / laptops to linux servers required to support the consumer internet. I know neither of those things are particularly well defined but it adds another perspective on this which’s be interesting.

It's not because Linux got better. It's because Windows got worse.

I dispute this notion. The truth is that "Linux" improved a lot over the years.

My first attempt at it was in 2002. At the time I was a kid that tinkered a bit with his own PC, not a technical user proper. I tried to install Mandrake Linux, and it was difficult to get it running. Ultimately I failed to get my mouse to work and gave up.

Many years later, at around 2015, I made a second attempt. My workplace supported both Windows and Linux, and I opted for Linux, as testing locally in an environment similar to what was found in Production was better. No shenanigans with line endings, casing on file names, etc. Ubuntu this time around. It was a joy to use. Installing it on the computer was extremely easy. I eventually dual-booted at my home computer too. My only real issue was that some software I wanted to use (games mostly) didn't run on Linux, which was a shame.

My dual boot setup lasted for a few years, but it was a hassle to switch in between both when I stopped coding and wanted to play games. And eventually I dropped it for Windows 10 + WSL.

Then I made my final attempt around a year ago. Linux Mint this time. I read how WINE/Proton had made huge strides into providing Windows compatibility, and thought it was time go give it another go. It has been an amazing experience. I can run the vast majority of my gaming library without any issues (really only one exception so far, a somewhat obscure game at that). After more than a year, there was not a single moment where I thought I needed to be on Windows for anything. And the OS is a joy to use. It has no right being this good for free. I even donated 20 bucks as appreciation. Will probably do it again this year.

So yeah. Things actually improved a lot. I agree that Windows has been getting worse, but that is not the full story

everytime i boot a Linux based disco on hardware for the first time im always blown away that some of the greatest multitools of mankind are free and available in minutes

Frankly, the chart looks like Linux's market share is below the range of errors of the measurments made.

Compared to Linux, the famous "Unknown" OS has a market share of 10%-ish, a spectacular come back from its steep decline between May and June 2023 where it went from 12% to 3%. Windows and OSX market shares aren't stable at all and can move 4% in one month.

So Linux going from 3.9% to 4% should be taken with a grain of salt...

I switched from MacOS to Arch Linux with KDE Plasma 3 years ago and couldn’t be happier. It’s so nice to have Pacman (and the AUR) as a built-in package manager. My experience with multiple monitors is better than on MacOS. My experience with gaming is surprisingly better than on Windows. Elden Ring worked day one on Linux but was almost unplayable on Windows for me. The entire experience feels to the point and working for me. Windows was littered with ads and weird driver issues. MacOS would constantly yell at me for trying to do things on my own computer. On Linux, I could simply plug in a printer and print documents. Finally, life as a developer is so much higher quality all around. Linux just works. Never thought I would see the day!

If one considers how desperately Microsoft tries to squeeze additional revenue from Windows (forced cloud account, ads, browser war 2.0) one has to wonder at what point they'll simply shut down OS (i.e., kernel and plumbing-level user-space tools) development completely.

I mean yes, it has been a proud piece of Microsoft but the OG NT kernel team probably left by now and from a management perspective it doesn't offer any revenue to develop an OS kernel.

Otoh, how big is their OS engineering team? Drivers are usually developed elsewhere. So maybe 100 folks or so? With a relatively low quota of non-devs, I assume? So Microsoft would potentially save what, tens of millions of dollars per year? Maybe they just continue this as some kind of tradition department.

Did you miss the 72% market share of Windows on desktop machines? That is still a massive market, and if nothing else, pulls consumers into the rest of the MS ecosystem.

So I doubt MS is abandoning OS development any time soon.

I don't feel like MS ecosystem is that attractive to consumers. In enterprise for sure, everyone uses Teams, Outlook and MS Office, and Azure is popular for hosting. But in most people's personal lives it's smartphones that are most utilized, which naturally draws people to Google's and Apple's ecosystems.

Only reasons I even bother to install Windows are gaming and MS Office provided by my university.

> how big is their OS engineering team? (…) maybe 100 folks or so?

From https://devblogs.microsoft.com/bharry/the-largest-git-repo-o... :

> the Windows team is about 4,000 engineers and the engineering system produces 1,760 daily “lab builds” across 440 branches in addition to thousands of pull request validation builds

Only a small minority of those people are going to be working on the core OS under discussion here.

Ah right, the gp comment was specifically about the kernel, not the whole OS, sorry.

Regarding the kernel, I think it's not really possible for Microsoft to port Windows to the Linux kernel without breaking its backward-compatibility policy. Or maybe it would need something like WSL but reversed? But anyway I'm not sure it would make much sense.

That said, I didn't even touch a Windows system for maybe more than 10 years now…

Microsoft notoriously has major issues implementing new features or fixing bugs due to their commitment to backwards compatibility.

E.g. I've been told personally by Microsoft developers in the know that certain oddities or performance differences between POSIX and Windows come down to a handful of obscure but tricky test regression failures, and nobody could justify the time spent on bridging the gap. This was in the context of improving Git performance on Windows.

So even if Microsoft made the management decision tomorrow to ditch NT and ship Windows as a Linux distro, it's unclear how they could square that with backwards compatibility in a way that would result in overall cost savings for the foreseeable future.

The weight of that single word "tricky" in tricky test regression failures is very high!

I can't imagine the obscure craziness you'll be encountering when working on a codebase that's easily over 25 years old, supporting billions and billions of devices.

Yes, although having never been at Microsoft (and thus just speculating) I wouldn't underestimate the managerial effect of siloing either.

There are many tricky problems in major companies that are actually relatively easy with say a month of developer time, but any single team affected by it can't justify that over a workaround that takes a week to implement.

Then nobody adds up cost of 10 different teams needing to each repeat that twice.

They are taking steps in that direction with Windows subsystem for Linux

I wonder if this is because fewer desktops are being bought (in favour of laptops), and those who remain on desktop are disproportionately Linux users.

Largest desktop market is china.

i saw folks suggesting a lot of the growth came from india, but i didn't look into it.

I believe Linux marketshare in India is somewhere around 15%?

I think the categories are "desktop", "mobile", and "server". Perhaps "tablet" as well.

So "desktop OS" likely includes actual desktops and laptops.

I would assume that laptops count as desktop because they are running desktop operating systems. How would a browser user agent string even differentiate between a desktop PC and laptop?

Laptops are desktops.

Many people never move their laptop from their desk.

More likely it's because fewer desktops AND laptops are being bought in favor of phones and tablets (and Chromebooks).

Windows appeals to two groups with any real loyalty: gamers, and enterprises. Everyone else who was just a casual user just used it for decades because it's what came preinstalled on the most systems. If all you're doing is watching netflix, checking social media, and reading email, you really don't care what OS you're using, unless it forces you to care (at which point, you consider another device).

Gamer loyalty is getting more and more dubious. I wouldn't necessarily say this means that we'll have a ton of Linux desktops though. We'll just likely be seeing fewer PC's in general, as we already have been. And those of us using Linux will be the last to ditch the form factor, as phones, tablets, and smart tv's are more of an appliance than a tool.

Reply on news.ycombinator.com