Meh, GNOME has 1/10th of the features of KDE, but it's much more stable and consistent.
I've used KDE for the past year, and it's just too much, too many options, and if you stray out of the happy path, you encounter plenty of bugs. Then what's the point of offering so many options. I'm back to GNOME.
KDE enjoys a lot of reputation from people that believe the Windows-style UI paradigm to be the best. That's arguable. I would certainly install KDE to a user new to Linux, but I have been running Linux long enough not to get lost if I don't have a taskbar or desktop icons.
GNOME could be so much better, sure, but I prefer 2 options that work (4 code paths to test), than 10 that don't really work all that well (1024 code paths to test).
My dream DE has the simplicity and design of GNOME with the completeness of QT. GTK is a dead-end, but at least it's written in C, so it is future-proof compatible with better languages such as Rust, instead of being stuck with C++ until the heat death of the universe.
> if you stray out of the happy path, you encounter plenty of bugs
But to me the happy path (the defaults) out-of-the-box on KDE are just better. The console and text editor are legitimately 10x better than GNOME's. The settings app, disk manager, the open/save dialogs, and -- especially -- the file manager.
I do most of my work in VS Code and web browsers, so I am not even a heavy user of the apps that come with the desktop environment, but the quality of those ancillary tools really dictates the quality of life in a GUI environment.
I ended up using GNOME a bunch in the last year because I have to use Wayland (X11 doesn't support my monitor setup) but remote desktop is an important tool in my day-to-day, and for a while only GNOME had a decent RDP story (for accessing the Linux desktop environment from Windows or Mac) on Wayland.
I think that is no longer the case, though, with krdp[1] — seems to have not made it into Plasma 6 after all, but it does work pretty well so far — so I am so excited for KDE 6 that I enabled the testing repos so I could install it on my Arch Linux workstation right away, without waiting for the official packages.
Definitely true (and I do install Konsole on GNOME if I have to use GNOME) but probably not super common.
Most people, myself included, are gonna install the DE and its apps by choosing it in the OS installer (or at least with a single command, a la "pacman -S plasma-meta kde-applications-meta sddm").
Stability is a mixed bag on GNOME. It's been a couple years but I was surprised last time I used GNOME to have Mutter crash back to gdm randomly while drawing due to a bug in graphics tablet code. I typically use SwayWM and while the graphics tablet support is nothing to write home about... It's very uncommon for it to segfault for me. My sessions in Sway tend to last months long, normally interrupted by rebooting for kernel updates or something like that. I do like that it can be extended with JS but that also ran me into all sorts of weird problems, more than it used to when GNOME was newer; I just want basic features like tray icons/app indicators...
(P.S.: I think I am probably the main user of graphics tablets in SwayWM, but if anyone had been using it, I'm sorry for the tool buttons being buggy in 1.8. It was my bug and it should be fixed in 1.9, fingers crossed, it looks like 1.9 will be hitting nixos-unstable later today for me to check.)
I have to periodically restart my session if I'm using Gnome with Wayland, as memory use keeps growing. With the X11 version, you could alt + f2, then "r" to restart gnome-shell. This is, for some reason, not possible when using Wayland.
That's because what is restarting, if I understand correctly, is Mutter. And under X11, Mutter is effectively an X11 client. But, under Wayland, Mutter is the compositor... it of course does still do compositing under X11, but under Wayland the compositor is also the display server. So you can't restart it without disconnecting all of the clients... kind of.
Crash recovery and graceful restarts of the compositor are things that should be possible and are being worked on, and ideally this will allow for well-written Wayland compositors to tolerate a variety of issues that would've been hard to on X11, but for now, Wayland compositors mostly can't be restarted. This is also why GNOME doesn't want too much complex stuff going on directly in the compositor, and can explain some other architectural decisions about GNOME Wayland that are otherwise peculiar.
I suspect that it's the appindicator extension that I am using which causes the problem, but I've not proven this. I'm still salty that they removed appindicator support to begin with, though.
To be completely pedantic, I don't believe the Wayland protocol itself actually dictates a design like this: you can separate the Wayland server from the compositor and display server bits if you want. I am not aware of many implementations of this, though; the best example is probably still Arcan.
That said, the very vast majority of Wayland compositors, including Mutter, Weston and everything using wlroots, is implemented without separation between the display server, compositor, etc. so in practice this is still mostly true, it just needn't remain true into the future.
You're right, of course, and I should've been more precise about that given I have looked at doing exactly that myself (main thing stopping me: I was able to switch to my own X11 window manager within a day - it was painful but worked; meanwhile I'd locked up my machine's display hard within 5 minutes of running some DRI/GBM test code and had to reboot)
I do think, ironically, that the future of Wayland will involve making it more X-like - adding WM support, maybe stripping back the exceedingly overcomplicated protocol (my window manager is smaller than most Wayland example clients..)
And thanks to the extensibility of the Wayland protocol, you can layer any X functionality right back in...
That's why I'm happy the KDE developers and others have acknowledged this is actually a problem and are creating solutions for it, unlike many GNOME developers who say "it's your fault it crashed!"
Yeah similar experience here, At work we are forced to use a distro with GNOME (well at least it is GNU/Linux and not that Microsoft bloated spyware) and yeah I have plenty of crashes in GNOME. No crashes at home with KDE Plasma on openSUSE Tumbleweed. It has been rock stable.
My personal KDE looks and operates nothing like Windows and more copies the MacOS workflow (although I am not a Mac user at all). GNOME is not that much customizable and it is the main reason I stick to KDE. Also, quite stable. I do rarely have any issues to be honest and it usually is Latte that has bugs but it is in the state maintaining limbo for a while now.
I've used KDE for the past year, and it's just too much, too many options, and if you stray out of the happy path, you encounter plenty of bugs. Then what's the point of offering so many options. I'm back to GNOME.
KDE enjoys a lot of reputation from people that believe the Windows-style UI paradigm to be the best. That's arguable. I would certainly install KDE to a user new to Linux, but I have been running Linux long enough not to get lost if I don't have a taskbar or desktop icons.
GNOME could be so much better, sure, but I prefer 2 options that work (4 code paths to test), than 10 that don't really work all that well (1024 code paths to test).
My dream DE has the simplicity and design of GNOME with the completeness of QT. GTK is a dead-end, but at least it's written in C, so it is future-proof compatible with better languages such as Rust, instead of being stuck with C++ until the heat death of the universe.