So first of all, this is not a “help me like linux” post but desktop linux specifically and it’s not a “linux is shit” post either.
I run a whole bunch of linux servers (including the one that hosts the instance I’m posting from), the first thing I install on a Windows machine is WSL and I’ve compiled my first kernel about 20 years ago so that’s not the problem we’re facing here. I understand how linux works and considering the end of support for Windows 10 this is as good an opportunity as ever to fully make the switch.
My problem is more that specifically linux on a desktop still feels more like an unfinished prototype than like something I’d want to use as a daily driver. About once a year I challenge myself to try it for a while and see how it feels. I look around for a distro that seems promising, put it on a spare SSD, put it either into my Framework laptop or my gaming machine and see where the journey takes me, only booting Windows in an emergency.
And each time, I get fed up after a few days:
- Navigating a combination of the distro’s native package manager (apt, pacman, rpm, whatever), snap, flatpack and still having to set up the maintainers’ custom repositories to get stuff that’s even remotely up-to-date somehow feels even messier than the Windows approach of downloading binaries manually.
- The different UI toolkits, desktop environment, window manager and compositor seem to be fighting each other. I feel like even for something simple as changing a theme or the UI scaling, I have to change settings in three different places just to notice that half the applications still ignore them and my login screen renders in the top left corner of the screen but the mouse cursor acts as if the whole screen was used.
- All of that seems to be getting worse when fractional scaling is involved which is a must for the 2256x1504 screen in my Framework 13.
- The general advice seems to be “just wait until you run into a problem, then research how to solve it”. For my server stuff, this works really well. But for desktop linux, it feels like for every problem I find five different solutions where each of them assumes an entirely different technology stack and if mine is even slightly different I eventually run into a step where a config file is not where it should be or a package is not available for what I’m using.
- I do a lot of .NET programming and photo editing. I could probably replace VS with VScode or Ryder but it’s an additional hurdle. For photo editing, I haven’t found a single thing that fits my workflow the way Bridge, Camera Raw and Photoshop do. I’ve tried Gimp, Krita, Darktable, RawTherapee and probably a couple more and they all felt like they were missing half the features or suffer from the same unintuitive UI/UX that Blender had before they completely overhauled it with 2.8.
Sooo… where do I go from this? I really want this to work out.
If you need theming stick with KDE and Qt apps. Gnome considers large amounts of theme to be a hack. Remember in windows there is no themeing so by switching that’s an extra feature you get. Even if half baked
In reality a lot of these desktop standards are competing. Some are working together but some clash. It’s part of the open source model
A general rule of thumb I follow is that if I’m getting a lot of resistance trying to do something, it probably wasn’t designed to do that.
Fractional scaling works better on KDE from what I hear to but I have no idea.
Personal something that has really helped me was switching to Nixos and configuring everything in a nix file. I don’t even complicate it with home manager just plane nix
Nix also makes installing developer binaries a breeze. Number one nix packages as the largest number of packages over any other repo. Two, it provides a very organized method for installing packages
Really good post and as a daily linux user for quite some time, I can relate.
I also strive for simplicity so I try to minimize the use of anything which isn’t in the official repos of a distribution. This is a big one, because depending on your use-case this trims down the applicable distros to your specific needs quite a bit. e.g. do you need up-to-date packages/firmware/drivers, you have to look at rolling-release distros. Do you not have these requirements and looking for stability then you can look at point releases (LTS) on the other side of the spectrum.
What most linux enthusiast forget to take into consideration is the willingness to use the command line. If you are very reluctant to use a terminal you should look e.h. for distros which give you the ability to configure almost everything through built-in GUIs e.g. OpenSuse’s Yast (Leap, Tumbleweed etc.) or Gnome focused distros (Ubuntu etc.) This also links into trying out the different package managers and their commands, some of them will seem more straightforward (apt install, zypper install) whereas others are more obfuscated (pacman -S, xbps-install). Searching for packages (apt search, zypper search, pacman -Ss/-Qs whatever, xbps-query)
In terms of desktop environments/window managers I also strive for simplicity, but I am willing to invest some time to setup the initial configuration. If you want something which works out of the box you need to try out different ones. KDE is a sure hit on most of those use-cases (esp. gaming, being very customizable although I’d recommend to not fiddle too much, similar workflow to Windows) if you feel comfortable with the workflow of Gnome, this might also be something for you (kind of tablet/phone). Otherwise go into the rabbit-hole of smaller WMs like openbox, labwc or if you are using terminal applications a lot, twms, i3, riverwm etc.
I think you should just run a windows VM with hardware graphical acceleration and run your windows specific apps Lee the Adobe suite and VS there. I use desktop Linux but I don’t really have many needs.
I really want this to work out.
Why?
I don’t ask that to talk you out of it. I like desktop Linux. I’m typing this on desktop Linux. I’ve been using desktop Linux for most of my adult life. I ask because your reasons will inform the advice people can give you.
I do a lot of .NET programming and photo editing [with Windows-specific proprietary software]
There isn’t necessarily a good solution to this. Those are large, complicated programs with very deep workflows that are almost certainly going to be dissimilar in any substitute software, which is itself going to be large and complicated with its own ways of doing things. Using those specific programs may be more important to you than what OS you run them on.
It looks like Photoshop is probably usable with Wine, while Visual Studio isn’t. Using Wine means putting up with occasional instability and reduced performance. If you spend a lot of time in Photoshop, this may not be for you.
Another option is to run Windows in a VM for those apps. This will likely work smoothly with regard to the apps themselves, and generally performs near native, but does mean a less polished interaction with the rest of your desktop.
Regarding Photoshop in Wine: unfortunately, missing GPU support is probably a no-go when dealing with 6000x4000 pixel, 14 bits per channel raw photos.
Also a tiny bit amusing that within 24 hours, it was rated it both “Garbage: It launches, that about it.” and “Silver: it pretty much works well with a few caveats.”
Why?
I don’t ask that to talk you out of it. I like desktop Linux. I’m typing this on desktop Linux. I’ve been using desktop Linux for most of my adult life. I ask because your reasons will inform the advice people can give you.
Because enshittification is becoming more and more unbearable. So far, Windows 10 (and to some extent even Windows 11) works for me but it’s getting worse and worse every year. I have no interest in OneDrive, Copilot, Recall and whatever MS wants to sell me next. I’d rather have a system that does exactly what I need, nothing more, nothing less. On servers and embedded systems, linux has done a great job for me over the last 20 (!) years.
Another option is to run Windows in a VM for those apps.
Kind of defeats the purpose if I run my two most-used applications in a Windows VM, doesn’t it?
I’m more than open for using something different and learning a different workflow as long as I can eventually get to the point where I can get roughly the same results in roughly the same time. I like tinkering with stuff and I’m not even opposed to write my own tools for closing a few gaps in the applications I use (see my recently started immichtools). But there is a limit. I just can’t afford to spend half my day working around problems that I wouldn’t have had on Windows.
Edit: formatting
Because enshittification…
Let me try to help you in a more psychological way: try focusing on how much more important those issues of privacy, respecting user self determination, etc are than all those little trade offs (which sound to me as much like a cranky resistance to change as anything else). You are going to have to accept changes you don’t like along with changes that you might eventually see as improvements if you give it a chance. But even if not, the enshittification element should outweigh all that.
I gave up Windows for Plasma 4.something, over a decade ago to avoid the enshittification of Windows 10, and even then I felt like it was a user interface improvement and it was painful going back every time I booted my window partition. I can’t even imagine how someone can put up with the shit Windows 11 imposes on you. But, hey everyone weighs things differently.
Personally, when software I paid nothing for, made by volunteers, has a flaw or doesn’t meet my preferences, it pisses me off a whole lot less than when software that I’ve paid for, made by a corporation with more money then God, blue screens or forces something on me that I didn’t ask for.
Kind of defeats the purpose
That’s why the why matters. Some people might just not trust Windows to keep private data secure, but be comfortable running certain software on it in a VM, possibly a VM that isn’t usually allowed network access.
If you’re sufficiently motivated to get off Windows to invest time learning different workflows, there certainly are options. It sounds like you’ve tried some for image processing and found gaps. People might be able to help fill them if you go into detail about your current workflow, but there is no 1:1 replacement for Photoshop on any platform. If you’re a heavy Photoshop user, there may be no path to happiness for you.
There’s surely a 1:1 replacement for Visual Studio outside of Windows-specific development (which wouldn’t make much sense to attempt on Linux anyway).
I think that this is just the reality. Linux is not Windows, and each has their issues. I run an Arch based distro so that I can get new changes as soon as is reasonable without needing to go to user repositories, but some things still lag behind Windows because there is no corporate motivation ($) to build those features for Linux.
I’ve changed my workflow and mindset about what I need to do and how to get there in Linux. This involves making tradeoffs when Linux can’t do things the same way, or at all. I think this is a common experience.
My earbuds’ firmware can only be updated with a Windows app. My tax software only works in Windows. I have to use a Windows VM for those.
There are online games I can’t play with friends. I use FreeCAD instead of Autodesk stuff, I made sure that I can use the slicer for my 3D printer in Linux before buying it, and so on.
Realistically, I can’t be the change that I want to see, so I live with it. For me, it’s still worth it.
So, I made the full switch to Linux about a year ago.
My journey has lead me down the Arch rabbit hole. And I feel KDE has the most complete feel. And you can make it as close to what you’re used to with Windows. Hotkeys included. KDE is improving vastly. I have no issues with fractional scaling on KDE backed by Wayland.
As far as package management goes. Sounds like you’re comfortable in CLI, so between Pacman and Yay package management is fairly simple. I have an alias “yeet” to uninstall.
As far as for coding, you can use VSCodium
I’m quite happy with Arch, and KDE on Wayland
So what are you using on the desktop and how long have you been using it? I’ve switched to Linux / open source software a year ago and especially learning programs (design / media tools in my case) was super painful for some time. But I had the motivation of not wanting to use an OS that tries to spy on me and getting away from a super shady monopolist (Adobe) as a motivator that kept me going. Sure there are still some things that get on my nerves but those exist in Microsoft’s (and Apple’s) and Adobe’s products as well.
I’ve been using it for quite some time now and I don’t see the issue. I mostly use Gnome and that’s kind of polished and minimalistic(?) looks very cohesive to me. But I believe the same applies to other desktop environments as well. My package manager mostly gets out of the way and I don’t have to pay too much attention to that. I even get browser extensions and all the stuff that ties into another from one and the same distro maintainers. I’ve tried other operating systems as well, but for the other ones I needed to install 50 small utilities to make it usable and those kind of fight each other as well. On Linux, I try to avoid Flatpak and I wouldn’t use Snap at all. We still(?) have most software available as proper packages.
I can see how image editing might be an issue. We have what we have and for the rest you need to get one of the commercial products running.
Re: packages
What software specifically are you not finding in the repos? If it’s FOSS there should be a simple way of getting it. If it’s proprietary you kind need to pick your least hated option:
- Third party repos - very hit and miss, easiest to troubleshoot, can be awesome rock solid or buggy as hell
- nix package manager - i have no experience but very popular right now
- Flatpak - can be managed by Gnome Software or KDE Discover, no system theme integration, config files are in weird places
- Snap - similar to Flatpak on Ubuntu, YMMV elsewhere
- Appimage - if no other option, similar to Windows binaries, cant self-update
I’d pick one and stick with it as much as possible. Mixing several solutions is where things get confusing (for me at least)
Re: settings
UI scaling is a rough edge on Linux. Non-integer scaling (1.25, 1.5, 1.75, etc) doesnt always give perfect results on X11, and Wayland scaling only works on Plasma, Gnome, and the various compositors. Themeing isnt really a thing on Gnome, so only Plasma has both good scaling and themeing, and Plasma is especially guilty of the “settings in 3 places” phenomenon. If you want simple menus and good themeing you can use MATE, Cinnamon, or XFCE but then you lose wayland scaling.
I have run into the same bug with Display Manager not scaling on Plasma, but i dont have this issue on Gnome or on x11 desktops. Plasma may be the common denominator here.
Re: laptop
My advice is to try to like Gnome. It’s got the best scaling support right now. Disable all extensions, learn the keyboard shortcuts for window and desktop management, learn the touchpad swipe gestures, and pretend you dont miss themeing. If you can get past the initial apprehension, you might find a modern desktop with an elegant workflow. Add back in as few extensions as you can live with; each one is a potential source of instability/bugs but as long as you keep it to a small number you should be fine. I have 4 extensions currently enabled, and i wouldnt go too much higher.
If you end up hating it then Plasma is the next best option for your hardware. Plasma is in heavy development and still has lots of small issues, but things should improve over time
or you could try Hyprland
Re: photo editing
Curious about your workflow. I do a lot of wildlife photography as a hobby and I find just Darktable to be too much. I usually end up cropping, adjusting brightness and colors, and then exporting to a jpg.
What sorts of things are you doing with your photos? I dont think i have a solution for you, just curious. Also, can you run an older version of Blender? There might be a containerized solution for that already.
Thanks for the long reply, lots of stuff to unpack here that I might have to come back to later. Might be helpful.
So for now, let me focus on your question about my photography workflow. I mostly do event photography (discos, concerts, conventions) but also occasionally studio and travel stuff.
When I come home from a shoot, I copy my photos to a network drive on my home server (running Ubuntu) which automatically gets backed up to an off-site NAS. As a first step, I use Bridge to label which photos I want to edit for myself, which for a potential client, which not at all. Nothing special, just running through all RAWs and marking them with star or color labels. For the editing step itself, I start out with Camera Raw. First an overall pass with lens correction, cropping/straightening, brightness adjustments (exposure, contrast, blacks/darks/lights/whites), white balance, dehaze, curves, whatever the photo needs. Then, depending on the subjact, a more in depth pass with spot removal and masked adjustments. Automatic subject masking has been a great time saver. If I need to go even more in depth (usually only for photos that go to an exhibition), I start editing in photoshop. As a last step, I use Photoshop Image processor to bulk export JPGs in the needed size and quality, optionally with a watermark.
(for those familiar with Adobe’s tools, you might be wondering why I don’t use Lightroom instead. In the past I’ve had problems with accessing the same library from different machines. This could probably be fixed but my current setup works fine so I never bothered)
For long term library management, I run immich on my home server which lets me tag and filter my photos as much as I want.
As for the Blender thing, I think I phrased that weirdly. It was not related to a specific problem or my photo editing process. It was just an example for a piece of software that started out with horrible developer-user UI and got a lot better when they completely redid the UI in 2.8.
Thanks for the detailed explanation! Very interesting
I second @Hellmo_Luciferrari@lemm.ee 's suggestion of Arch/KDE. I cannot think of a time when I have had to go outside of pacman/yay or setup a custom repo. I don’t use them personally, but the AUR even has -git versions of many packages if I needed to have the absolutely down-to-the-minute latest version of something, but even the official repos are rarely far behind in my experience. Fractional scaling seems to work fine on my 2560x1600 display. VScodium works fine when I tested it, but I do not use it regularly enough to really have a strong opinion on it.
Just to put some attention on it, custom repos are AT LEAST on similar level than using yay for AUR packages. AUR is Arch’s biggest hurdle imo.
How so? I have never found installing yay difficult, and using it is essentially the same as pacman with the addition of aur packages. What issues have you run into?
AUR is not to be compared to arch packages.
Warning: AUR packages are user-produced content. These PKGBUILDs are completely unofficial and have not been thoroughly vetted. Any use of the provided files is at your own risk.-
https://wiki.archlinux.org/title/Arch_User_Repository
And exactly this is the reason it is its biggest flaw. People don’t read what AUR actually is.
I don’t do a lot of coding, but to write out any of my scripts or config files; VScodium works well enough for me.
2560x1440 and fractional scaling works for me. And even on my odd resolution 2240x1400 on my laptop it works well.
Using Pacman and Yay for AUR I have never had to dabble in adding other repos. AUR usually has anything I could want. When it doesn’t, I build it myself from source.
I’ve tried KDE, and Gnome, as well as many other desktop interfaces; but KDE always happens to be the one I go back to. Hell, I’ve even dabbled in Hyprland. I can’t say I love or hate it; it’s just got a learning curve.
But the fact that I can have many desktop environments/window ma ager installed and switch is a beautiful thing.
I use Windows for work and Linux (Debian/Plasma) at home.
I literally have none of these problems.
Cool. Can you help me solve mine so I don’t have them either?
Navigating a combination of the distro’s native package manager (apt, pacman, rpm, whatever), snap, flatpack and still having to set up the maintainers’ custom repositories to get stuff that’s even remotely up-to-date somehow
This sounds like a you problem, to be honest. If you want the most up-to-date software, just use a distribution that updates very often or uses a rolling-release concept.
The different UI toolkits, desktop environment, window manager and compositor seem to be fighting each other.
If you use one of them, not that much. If you start mixing them it becomes a huge mess. At one point in time I had Ubuntu installed, running Gnome, but having Openbox as window manager set. It was an absolute mess. Nowadays I think it’s even more of a mess, especially with gnome and this stupid Adwaita library with the stupid CSM.
But I happily ran pure Openbox on X11 for a decade and run labwc on Wayland since ca. 2 years now.
I do a lot of .NET programming and photo editing. I could probably replace VS with VScode or Ryder but it’s an additional hurdle. For photo editing, I haven’t found a single thing that fits my workflow the way Bridge, Camera Raw and Photoshop do.
Then stick with Windows. Or run this software in VM with GPU pass-through and KVM. I really don’t see an issue here. Use the tool that best fits your needs.
I don’t think desktop Linux works for you. That’s the truth. You know, I was like you, and so was my husband. These individual projects that never felt that they fit together to create a cohesive product, always bothered me. So what you’re asking, will never get fixed, to be honest with you.
But speaking about myself, I decided to use Linux because it’s the right thing to do. As a painter myself, who needs some of the features Photoshop has but Gimp 3 doesn’t, I feel you. But still, I use Gimp now, 100% of the time. I settled for less, because again, it’s the right thing to do. I have no interest to use Windows and its spyware. I have a macbook air with macos for occasional browsing (I like the hardware), but again, I use OSS software on it (including gimp). The rest of my 5 laptops and 3 desktops, all run desktop Linux. I’m more often on an old Macbook Air from 2015 running Linux Mint, than I am on the new Mac running MacOS. My main desktop is Debian-Testing. Is it as cohesive as Windows? No, and it will never be. But again, it’s the right choice.
I guess you have checked out Krita? I like it a lot.
The right choice doesn’t help me if I can’t get my literal job done and have to give up half of my hobbies.
Thankfully for me, I don’t work anymore, so it’s easier to move to Linux. But my hobby (painting), it’s done in a less optimal degree with Gimp, yes. And that’s ok with me.
I’ve been “Linux-adjacent” for years, and recently switched my main gaming computer over to it. And I’ve seen exactly those frustrations so many times.
The good AND bad part about user-managed software is that the developer-users decide how things work, then things stay that way until other developer-users do things differently.
My most recent frustration? Drive automounting on boot.
On Windows or Mac, all physical drives mount when the system boots up.
On most, but all, varieties of Linux, it seems ONLY the system drive is mounted.
This gave me trouble when I tried to set a second drive as the default location for Steam.
Every time I rebooted, the Steam client forgot that I had a second hard drive. I didn’t realize why, because in system settings I told the computer to mount all drives on boot.
But. But.
By default, Bazzite seems to set secondary drives as external, rather than internal. Spork knows why.
So I had to sift through forum posts until I discovered that the internal drive was being seen as external. Then I had to figure out the combination of partition management tools and console commands to tell the system to mount the drive as an internal drive, rather than external.
It now works perfectly - after over an hour of research and a couple days of frustration.
There are two problems: 1. An extremely basic thing doesn’t work the way the majority of users expect it to, and 2. A majority of developer-users apparently think it works fine as it is and doesn’t need to be changed.
So I feel your pain. I’d rather be using Linux now for gaming and for my 3D printing related hobbies.
But for my day job, I’m on PC or Mac. I have to be, because I can’t stop working for two hours while I troubleshoot and find a solution to an obscure problem.
I started using Linux every day in 1999 and I’m glad I did.
Managing a Linux server is no different from managing a Linux desktop. If you were to consider the GUI nothing more than a display layer over the top of a server, you’d have a good mental map of how things work.
To get started, use the same desktop distro as your server and use their preferred or default windowing system.
Once you’ve familiar with it and the pitfalls it comes with, you’ll know which questions to ask for your next choice, but you will be able to build on what you already know.
What you’re proposing is exactly how I got to the point where I’m writing this post. My servers are mostly Ubuntu, apart from a couple of Pis that run Debian. So naturally, I’ve tried Ubuntu, Mint and Pop!_OS. I can’t remember exactly which desktop environments I’ve tried over the years but at least Gnome, KDE, XFCE and Cosmic. Probably more.
When that didn’t work out, I tried Fedora and even some Arch-based distro (I think it was EndeavourOS).
Each time I ran into the same frustrations. Stuff didn’t work and troubleshooting consisted more of filtering which guides are actually applicable to my current combination of software than actually solving the problem.
Each time I ran into the same frustrations. Stuff didn’t work and troubleshooting consisted more of filtering which guides are actually applicable to my current combination of software than actually solving the problem
I think herein lies your problem, you need the ability to adapt problem solutions aimed at other software/setups at your own. I know it sounds easier than it is, but thats the gist of working with linux.
If you’re patient and want to gain a deeper understanding, try Arch itself rather than an Arch-based distribution that’s easy to install.
You’ll spend a long time on the initial installation and setup and you’ll read a lot of documentation in the process. When you have a usable system, you’ll understand what’s installed, how it’s configured, and why. Expect to spend a couple days just to get it usable though - this approach isn’t for everyone.
The Arch docs are top tier, but they’re not necessarily step by step guides because there’s more than one way you might choose to set things up. The docs tell you how the pieces can fit together, but it’s ultimately up to you to to do the assembly.
One of my first experiences with linux was gentoo back in ~2006 so patience is not an issue. Documentation that requires you to already know what you need to do is a problem though and the exact reason why I haven’t touched proper Arch so far.
Ubuntu sucks ass, and so do its derivitives. Arch is for people who still drive manual cars. I suggest sticking with fedora or debian.
