Giles Orr via Talk said on Fri, 26 Sep 2025 12:10:33 -0400
Tiling window managers have existed for a couple decades. (I said "window managers," because the "tiling" paradigm long predates Wayland.
Very true. And notice that the very same Dedicated Followers of Fashion salivating over the new tiling compositors wouldn't have been caught dead with X11 window managers awesome, dwm, ratpoison, or i3, because the tiling window manager concept was !!!OLD!!!. I mean really, really old: I was using a tiling window manager in 1987 on a VAX workstation.
And plural, because there were many of them by the time the Wayland project started in 2008.) They have tended to be very stripped down, as most of the people who use them are trying to lose the distractions of window decorations and window management - let the WM or compositor handle that. Likewise, why clutter things up with taskbars and stuff? But others have seen this and thought they wanted that simplicity ... but with extra features like launchers and multiple desktops and a good file manager etc. etc. This is both the wonder and the torture of Linux: "do one thing and do it well," which means that you can get the best of every category of software ... and you have to learn every single piece of your perfect desktop separately.
Tiling compositors also mean you have to learn keyboard controls, and that's not a paradigm that appeals to most people.
The Niri tiling compositor brings up a keystroke cheat sheet every time you open a new terminal emulator. This goes a long way toward eliminating the objection of having to memorize keystrokes.
Even though keyboard controls are almost always faster - but only if you can remember them.
The trick is to be prompted in your keyboard usage, until it becomes part of your muscle memory.
I'm happy you've discovered something exciting about Linux, but being a documenter of WMs and compositors, I'm too literal-minded to let that word "new" slide by. :-)
LOL, exactly! <my-opinion> Here's my opinion on the sudden repopularity of tiling user interfaces with Wayland compositors: It's relatively easy to make a window manager (WM) for X11. The WM tells X to establish an X style window (XSW). An XSW includes only the window content, not the decorations. X back-reports every XSW event to the window manager, which then handles WM type things like changing the titlebar color and font. Thankfully, X11 does most of the pixel arithmetic, so there's a separation of concerns, so creation of a new window manager is doable by one smart person. Wayland is a specification, not a piece of software like X11. So to create a compositor, you need to include most of the functionalities formerly handled by X11, plus the functionalities of a normal window manager. But wait, there's more: You have to be familiar with the entire Wayland *specification* to create a compositor. So, with all this work to do, is it any wonder that most of the compositors created by non-huge entities follow the simpler tiling layout? It's not that tiling suddenly became great: It's that overlapping was just too difficult to achieve for smaller project in the Wayland ecosystem. Wayland was a suboptimal solution to a very real problem. First of all, X11 itself should never have been a server. For those who simply must ssh into a GUI, there should have been a small, thin interface that made X available through a server. If X hadn't been a server, it would have eliminated most of the security concerns, and could have been run by a normal user, etc. Second, X11 has acquired way to much stuff, from what I hear. Like KDE and systemd, it has become an entangled mess in a black box. A lot of user requests should have been rejected. I think Wayland could have some potential moving forward. Somebody someday will create a Wayland-compliant module that performs the base functionalities performed by X11. I don't mean plug in compatible with X11, and I don't mean XWayland, I just mean encapsulating the part of current compositors that doesn't, or shouldn't, change from compositor to compositor. It should have a nice, simple, thin, non-dbus interface to the part of the compositor affecting user interface and user interaction. Somebody will also write a module to handle software interactions, so something like Tk, and as simple as Tk, can be used in software written for Wayland. Given that the module encapsulating the former functionalities of X11 would probably have most of the same functionionalities as X11 (nice sentence so far, huh?), it's possible Tk can still be used with only a minimum of changes. Here's the problem. These nice, neat, thin-interfaced modules that would have made life so much easier will almost certainly get hijacked by FreeDesktop.Org and its corporate buddies, and turned into hypercomplexificated dbus-laden monstrosities, perhaps even dependent on having a certain init system (guess which one). I'm hoping for a miracle. </my-opinion> SteveT Steve Litt http://444domains.com