> Linux desktop is getting better but I still wouldn't daily drive it,
I'm genuinely interested what Linux is missing for you? I've been daily driving it for years and do all my work and gaming on it. Is it specific software or?
I'm genuinely interested what Linux is missing for you? I've been daily driving it for years and do all my work and gaming on it. Is it specific software or?
- fractional scaling did not work in Gnome with Wayland for X11 Apps
- I still cannot use my LG C4 as a monitor in full capacity because AMD on Linux does not support HDMI 2.1
- Screen sharing was very buggy - in Slack especially - it would constantly crash the slack app during calls, ditto for camera, but even in Google meet and Chrome I've had desktop crashes
- When I switched to KDE/Plasma 5 to get fractional scaling it was extremely unstable
- Right now I upgraded my GPU to 9070XT - I'm still not sure if that would work on Linux yet because of driver support delay
- Guitar Amp simulator software I use does not support Linux, neither does Ableton (which supposedly can run on proton but with many glitches)
- The audio DAW situation was way too complicated and buggy
- I spent days to get the distro functional and usable with Ardour and it would still crash constantly - I just wanted to run some amp sims :(
It's just the little things and rough edges, but for example the fractional scaling stuff already improved because more apps that I use added Wayland support. And the emulation is getting better, with more users I could see larger DAWs supporting Linux as well. Not sure about the audio progress - JACK was a complete mess.
You can install AMDs driver from their repo directly, it works just fine (using it every day).
> I still cannot use my LG C4 as a monitor in full capacity because AMD on Linux does not support HDMI 2.1
That will never be possible. To prevent pirates from breaking it (lol), HDMI has decided to keep HDMI 2.1 secret. No open source version of HDMI 2.1 can exist.
That said, AMD's driver repo includes both the open source drivers and some proprietary versions of the driver, maybe that'll work for you.
Another option would be using a displayport output and a DP to HDMI converter, as e.g. Intel is using for their GPUs.
- HDMI 2.1: The HDMI Forum blocked it, as they don't want the details of HDMI 2.1 publically available. If you can, use DisplayPort, which is an actual open standard, and is better anyway. Nvidia works because they implemented it in closed-source firmware instead. https://www.phoronix.com/news/HDMI-2.1-OSS-Rejected
Strangely enough Plasma was able to handle this regardless (guess it was misreporting the resolution to X11 app or something like that to make it work ?) it was a Gnome/Wayland thing.
DisplayPort isn't an option - the TV only has HDMI in and converters suck (they crash constantly, even the expensive ones)
* The free trial is enforced as heavily as WinRAR's, and it's pretty cheap (~$60) to buy a licence if the nag screen makes you feel bad enough
KDE Plasma 6 made major improvements and has excellent fractional scaling, the best I've seen in a Linux desktop environment and comparable to scaling in Windows 10-11. I encourage you to give it a try.
and now I’m constantly getting these complaints “I can’t get screen capture to work under Wayland… I switched from lightdm to sddm and I can’t work out how to switch back… I accidentally started an i3 session and I can’t work out how to log out of it.”
It makes me kind of miss Windows, in a way. It is good he’s learning so much. But the downside is Linux gives him lots more ways to break things and then ask me to fix them for him. And a lot of this stuff I then have to learn myself before I can fix it, because most of my Linux experience is with using it as a server OS, where desktop environments aren’t even installed
It's harder as a parent to know that you're capable of solving their problem and still say no, but by age 12 that's pretty much your primary job: to find more and more things that they can start doing for themselves, express your confidence in them, and let them figure out how to adult bit by bit. Breaking a Linux install and fixing it again is among the lowest stakes ways that dynamic will play out from here on.
Well, there’s your problem ;-)
This is great, though, really. I broke our computer so many times growing up, I couldn’t possibly count. I don’t think I ever lost anything of import, other than some savegames of mine. I keep telling people who ask, “how do I learn Linux?” that they need to use it, tinker with it, break it, and fix it, ideally without anything other than man pages and distro docs. It is a shockingly effective way to learn how things work.
It isn't that he could do that, but what else to give up?
Examples (I've been on desktop Linux since 2009): shutdown actually reboots except for a few months with some lucky combination of kernel and nvidia driver. The brightness control keys didn't work for at least half of the years. They currently work. All of that has workarounds but I understand that some people legitimately fold and go using another OS.
Oh he'll figure it out eventually. This kid might be going places.
I started with Linux installing it from floppy disks in about 1996.
In 1995, I was back on Windows 95 within a week because I needed to get something done.
In 2000, I was back on Windows 2000 within a week because I needed to get something done.
In 2005, I was back on Windows XP within a week because I needed to get something done.
In 2012, I was back on Windows 7 within a week because I needed to get something done.
In 2015, I was back on macOS within a week because I needed to get something done.
In 2020, I worked out I'm wasting my time on this.
I watch my colleagues and friend struggling with it. Lots of small papercuts. Lots of weirdness. Lots of regressions. Plus many years of server-side experience says to me "I should probably just use FreeBSD" in that space.
It just worked in Linux. I don't get where this comes from, because every time I hit a problem in Linux, there's a solution.
In windows, you get a vague hex error code that leads you to a support page where the error could be caused by one of a dozen reasons.
And on top of that, MS is constantly hostile to any user who just wants a basic OS to use their computer with.
Secondly, there isn't always a solution in Linux. I've got one now where something is utterly broken and it's 5 layers of maintainers down and no one gives a shit.
I want to upgrade in order to retain that local account.
Then use massgrave hwid activator.
Steam getting proton was a godsend, all those years of games became playable so now I have a huge back catalog.
Not a bad idea. This is exactly what I do on my daily driver.
In windows, I can just shut the lid and not worry about it, because it will sleep first, and eventually hibernate. Ubuntu would just sleep until the battery dies.
I found instructions for enabling hibernate in Ubuntu, and they did make it show up in the power menu, but it didn't seem to work. (Which is presumably why it was hidden to begin with.)
I also tried NixOS, but I couldn't even get it to boot the installer.
It's really funny because this is one of the things I absolutely do not like about Windows. I absolutely hate it that I put the computer to sleep and when I come back the next day it has hibernated. That said, I agree that hibernation has always been finicky on Linux, however, I would say Ubuntu is not the best distro for this use case. I have been using Fedora and they even publish official guides for it[0] that's how seriously they take it.
0: https://fedoramagazine.org/update-on-hibernation-in-fedora-w...
I do this for arch Linux on my framework and it's fine. Startup time is under ten seconds, essentially zero battery drain, right back in your session with all apps/docs open.
Hibernate is definitely better but still finicky even on Mac/Windows, machines can and do fry themselves, or require a hard reset if you unplug a device at the wrong time. Or unexpectedly continue draining the battery.
It's a terrible, funky, poorly documented, exception filled world down in the low power states for hardware.
It won't come back up ok, as recently as 6 weeks ago.
Or the rampant reports of things like this: https://discussions.apple.com/thread/255642823?sortBy=rank
But too many companies have discovered that a docile "user" who's fed constant dopamine hits and has no actionable way to use a device other than open their wallet and fork over cash to watch more cats dance, or shop on more stores is exactly what they want.
Why don't you just click here and pay for Onedrive. Or just click there and accept Apple's new ridiculous terms.
If you just want to watch cats dance... you do you. I'll just keep doing me over here.
I once tried to set up a GPU passthrough setup to a Windows VM to play WoW but there were a ton of report that Blizzard just banned players for using QEMU VMs because they were marked as cheaters.
Primagean recently said that in a video commenting PewDiePie's "I switched to Linux" video. While he's apparently a good programmer (he worked at Netflix), he uses Vim, so I don't trust him. Edit: part about vim is an edgy joke.
I work in AAA gamedev and have deployed kernel level anti-cheats before, and I’m aware how unpopular they are; so, sorry for that… you would also accuse us of “bad programming” if there was an overabundance of cheaters that went undetected and/or uncorrected.
The answer is unfortunately complicated, the kernel level anti-cheats themselves aren’t necessarily poorly written, but what they are trying to do is poorly defined, so theres a temptation to put most of the logic into userland code and then share information with the kernel component- but then it’s dangerous for the same reason that crowdstrike was.
Not doing endpoint detection is also a problem because some amount of client trust is necessary for a good experience with low input latency. You get about 8ms in most cases to make a decision about what you will display to the user, that’s not enough time to round-trip to a server about if what is happening is ok or not. Movement in particular will feel extremely sluggish.
So, its a combination of kernel level code being harder in general (malloc, file access etc; are things the kernel gives you in user land after all), the problem space being relatively undefined (find errant software packages and memory manipulation), not being able to break out of the kernel level environment for an easier programming and iteration experience and trying to not affect performance.
Lots of people think they can do it better, I’m happy to hire anyone who actually thinks they have a clue, it’s a really hard problem honestly and the whole gamedev industry is itching for something better: even us gamedevs don’t like kernel level anti-cheat, it makes debugging harder for ourselves too and introduces hard to reproduce bugs.
PS; sorry if I’m not being eloquent, I am on vacation and typing from my phone.
However, what if Primeagen meant that HAVING to IMPLEMENT kernel level anti cheat is a symptom of bad programming, and not the anti cheat per se? (that is, with good enough programming, it could somehow be avoided).
And kudos to you. I appreciate people in game dev, they can get a lot done in short time. I haven't played mmo fps since battlefield 3, and it wasn't that bad then. But I've heard that without kernel level they would be unplayable.
Thank you for your time!
Long term I'm kinda hopeful that this is something that will be mitigated through AI-based approaches working to detect the resulting patterns rather than trying to detect the cheat code itself. But this requires sufficiently advanced models running very fast locally, and we're still far from that.
I don't understand, how could crowdstrike have avoided their issues by putting more code in the kernel? Or am I misreading your statement?
If they had not tried to parse data inside the kernel it would not have been an issue.
The reason is the round trip time mainly.
Server corrections will feel like “floaty” or “banding” behaviour, we used to do that and people get upset because it “feels” wrong.
About 40% of games that use anti cheat currently work on linux. Getting banned for using wine is very rare because anti cheat that don't support linux would complain about not running an prevent you from even joining a game to get banned.
I've been through enough KDE, QT, and Gnome API changes. It's just not where I want to burn my limited time.
My first GDI programs still compile.
OLE? Sure, let every application talk to the DLL components of every other application! What could go wrong? Data wants to be free right? Spread the love.
Making the desktop into a live webpage? And of course let any webpage happily load whatever binaries it wants from the internet. Super handy stuff. For some people more handy than others (really how this did not cause a mega Wannacry-event back in the day I don't understand)
There is a reason this stuff is legacy. The only reason it still compiles is because some companies have spent millions on custom developments 20 years ago that nobody remembers how it still works. Not because you should still be using it :)
> Remember MDI Multiple Document Interface? Having Windows within windows. It was a terrible idea.
It was definitely overused - nobody needs Microsoft word to be a window manager for every doc file. But it ends up growing into something really nice where you get to build out the sub windows of your IDE wherever you want them.
> OLE? Sure, let every application talk to the DLL components of every other application! What could go wrong? Data wants to be free right? Spread the love.
This was also incredible. I built one of the first tabbed web browsers by embedding instances of the IE 4 DLL into my tabs. OLE and OCX extended object-oriented programming across program and language boundaries.
> Making the desktop into a live webpage? And of course let any webpage happily load whatever binaries it wants from the internet. Super handy stuff. For some people more handy than others (really how this did not cause a mega Wannacry-event back in the day I don't understand)
This was terrific for pranks. Yeah what were they thinking on this one.
> There is a reason this stuff is legacy. The only reason it still compiles is because some companies have spent millions on custom developments 20 years ago that nobody remembers how it still works. Not because you should still be using it :)
Maybe as I've aged the novelty of building programs has worn off. For some things I don't want to have to port or even recompile it. I just want it to run. If Win32 is the only stable Linux ABI, GDI is the only stable Linux GUI toolkit.
And I know someones franticly typing away right now - yes, I am fully aware you can customise things, but out of the box it should be pretty damn well polished so that you don't need to.
Ubuntu's probably got the closest but it still just doesn't quite feel like they've nailed the experience.
It'd be interesting if there was a "Ubuntu v2" type effort, over 20 years later. Before ubuntu it's not as though desktop linux was an impossible dream or there was a lack of distros, but Canonical cleaned up a lot of rough edges to the extent it became a lingua franca. It's to the extent you can rely on ubuntu being in instructions for linux software, for example if there's any differences to required package names it'll be the ubuntu names over debian's.
1. My capture card doesn't work reliably in any distro. I'm not a gamer so I can't use a cheap and ubiquitous USB V4L card, I capture retro computing screens at weird resolutions and refresh rates so I have to use an enterprise-grade solution that can handle strange things like sync-on-green from 13w3 connectors and extremely rare outputs from UNIX workstations from the 80s and 90s.
2. If someone sends me a link on my phone it is difficult to copy and paste it to a Linux system.
3. Battery life on laptops, despite decades of improvements, is atrocious on Linux. If my laptop gets twelve hours of real-world use under OS A and six hours under OS B, I've got to use OS A.
4. All of my screens are 4K. Today, in 2025, a full decade after 4K became standard, the way various DE/WMs handle scaling is embarrassing.
5. Nvidia. Yeah, it "works" for about 2-3 kernel upgrades then you're greeted with a blinking cursor upon boot because of DKMS or some random reason like patching the system and not rebooting for a couple of days and then patching again.
6. There's little consistency across devices. When I log in to system A I want every single icon, file, and application to be the same as system B. iCloud/Onedrive do this. You can do this on Linux while on a LAN with remote home folders. I don't work exclusively on a LAN. Or I can set up puppet/ansible for my non-infrastructure systems and that makes me throw up in my mouth.
Almost none of that is the fault of the kernel. That's irrelevant.
For headless servers, I want nothing else. For a daily driver, as much as it pains me, nothing comes close to the Apple ecosystem. Apple Silicon is years ahead of everyone, and their interop with (admittedly only their own) other hardware is incredible. Universal Clipboard is magic. The fact that I can do nothing more than open an AirPod case and my phone registers it is magic. Finally, the fact that MacOS is *nix is absolutely icing on the cake.
To give a very concrete example, I have two identical Thinkpad T14 at work, one running Linux (Debian Bookworm with KDE) and one running Windows 11. When doing normal office work, the Linux laptop easily lasts a whole workday with >20% battery left at the end. The Windows laptop runs out of battery in less than 2 hours.
> Today, in 2025, a full decade after 4K became standard, the way various DE/WMs handle scaling is embarrassing.
Perhaps Syncthing would partially cover this? Not the applications, but the files ....
Also I really dislike how out of memory conditions just causes everything to grind to a halt for 5 minutes before something, typically Firefox, crashes. On Windows at least just Firefox gets very slow, but usually I can just nuke the process that eats too much memory. Not so on Linux as the whole desktop becomes unresponsive.
And every now and then I still need to fiddle with some config files or whatnot. Not game breaking but annoying.
I've listed some of which I encountered on Mint here https://www.virtualcuriosities.com/folders/273/usability-iss... Among them: AppImages just don't run unless you know how to make them run. This could be fixed with literally a single dialog box. There is no way to install fonts by default other than knowing where to put them and knowing how to get there. Every app that uses Alt+Click, e.g. for picking a color, won't work because that's bound by default by the DE.
These issues may sound small at first but think of it this way: did nobody making this OS think about how users were going to install fonts? Or ever used an application that used the Alt key? Or did they just assume everyone would know what to do when they download an appimage and double click on it and nothing happens?
And you can just feel that the whole thing is going to be like this. Every single time in the future you want to do something that isn't very extremely obvious, you'll find a hurdle.
I even had issues configuring my clock because somebody thought it was a good idea to just tell users to use a strftime code to format the taskbar clock. I actually had to type "%Y-%m-%d%n%H:%M" to get it to look the way I want. And this isn't an advanced setting. This is right clicking on the clock and clicking "Configure." When I realized what to do I actually laughed out loud because it felt like a joke. Fellas, only programmers know these codes. Make some GUIs for the normal people.
The fragmentation of Linux leads to a ping-pong of responsibilities. Linux can never be a bad OS because it isn't an OS.
On Windows, if the file manager is bad, that's Microsoft's fault. Period. Nobody tries to say "actually..." it's Microsoft's fault. Period. The same goes for the taskbar, for the control panel, for MS Paint, for even Microsoft Office. If Microsoft will fix it or make it worse depends on them, but nobody denies who is to blame and everyone know where the blame lies. Meanwhile I don't even know if the basic utilities that my distro distributes are under the responsibility of Mint's team or if they will just direct me to some random open source project's issue tracker if I start complaining about Celluloid or the "Drawing" app.
You can't talk about Linux thinking only about the good parts, or you aren't inviting people to try Linux, you're inviting them to try your distro. "Linux" means the whole ecosystem, including all of its problems.
Personally, I like modern Gnome: https://news.ycombinator.com/item?id=43859753
Personally, I find modern Gnome insufferable because it is non-customizable to the extent that even macOS only dreams of, and it doubles down on the modern trend of hiding important UI behind poorly discoverable gestures (active corners etc). Except their take on it is even worse in general for mouse users because of how much more "legwork" it adds - e.g. in a default Gnome setup on Fedora, you need to move mouse cursor in the top left corner for the dock to show up (so that you can switch apps or launch a new one)... but then it shows on the bottom of the screen, so now you need to move the cursor all the way there across the screen.
But that's all subjective and not really my point. The point, rather, is that Gnome looks and behaves very different from Win11 and macOS both, in ways that don't make it easy for users to migrate (and in fact they specifically state that their UX design does not consider that a goal).
I like they ditched all the unnecessary things from the settings. I think all the pro-level settings must be dealt with via terminal. That way, it’s both of two worlds. Me, I don’t mind it. But if I manage the computer for someone, I want them to have only the minimum things, so they won’t be overwhelmed. That’s very wise, and unfortunately all these Win3.1 geeks are complaining it’s bad. Yeah, okay, keep using your favourite XFCE then, or whatever.
I’d install Gnome for elderly, even if they have some previous Windows experience. Because they can afford to just ignore it. My mum, she has no computer, and last time she used Windows was like, idk, a decade ago. Explaining Gnome to her is easy: here is the Windows (or CMD) button, you press it once, you have this iPad like interface. Here is the Dock, you have all the necessary apps in there. More of them if you press that Windows button one more time. But actually you don’t need it 99% of the time, so you can survive with top left corner pressed once. Two times press is for me. Closing the app is that X button. What else does she need?
Now, try to explain the [any other DE basically] to elders the same way. Considering most of these people have iPads. And if they’re not, well, I don’t really get why, they should. My guess is that their interface appeal to that audience. And to me that’s a great thing, that’s most of non-tech people now.
However, I’m (being an obviously pro user) able to use the default Gnome productively. Almost as productively as I use SwayWM. To me, that’s very impressive.