> it puts macOS, built by Apple's army of engineers and dump trucks of money to shame

I don't really think they're designed for the same purpose or audience, so I don't think it shames the Apple engineers.

Well macOS has tried and failed, with their mix of "spaces", fullscreen apps, and maximized windows.

Edit: what I'm saying is they did try to do the fullscreen (and even tiled) modes.

> macOS has tried and failed

I, along with many millions of people, are using spaces and fullscreen apps daily with no issue. I have used i3 and sway before, and I much prefer the macOS UX. To say that it has failed is sheer arrogance.

On what do you base such a ridiculous claim?

I've never seen anybody use fullscreen-apps/spaces/maximized-windows smoothly, ever.

It's especially bad on multi-monitor.

  - New windows appear on unfocused monitor at random.
  - Can make shortcuts to spaces, but not the fullscreen-apps (which look like spaces to all other effects). There are apps to make shortcuts to apps, but this is useless if you have e.g. two terminals in fullscreen mode, it just targets one window at random.
  - Cannot deactivate the slow change-space aninmation.
  - Windows are grouped under same app in Alt-Tab list.
  - When fullscreen-app, some dialogs do overlay the app, some are hidden on some random space, and some forcefully move your active space. At random.
  - Plugging and unplugging monitors frequently wreaks havoc among the window disposition.

totally agree on maximized windows and spaces - I run 3 monitors, and use neither

You're missing the point. Sway is a team of people working on their own time out in the open to create something they really care about for almost no money. The final product is staggeringly good compared to a commercial OS like macOS, and yes, tiling WM requires more up front investment and isn't a normal pattern for most folks, but that shouldn't make people feel like Sway or tiling is beyond them.

But they aren't making an OS. They're making a window manager. Why are you comparing a WM to an OS?

Because I’m not being pedantic, and because it’s impossible to look at Sway outside the context of Linux.

Why do you want anyone to feel shame about there being another product that does things differently, though? They weren't trying to achieve what Sway did, and what they did achieve is great for what most of their users want. No need to try to put them down.

I really don't get the benefit of a tiling window manager. I tried one and instantly felt boxed in. There's not enough room on the screen for everything I need to have opened and flip between, which is why I use an overlapping window manager in the first place.

With a tiling window manager, every window you open has cognitive overload. Which workspace, which location on screen, is it okay if other windows resize, or do I have it float even? Every tile you add affects the other tiles in some way - at the very least a new tile (e.g. on a new workspace) consumes space that cannot be taken by other tiles. The reward for this additional cognitive load is that every action is concious - reducing a lot of the mess that can pile while using a comptuer. This makes them incredibly suitable for people whose computer use is predictable, organized and focused on quick parallel tasks. I understand why so many developers use them.

With a stacking window manager, the cognitive overhead of opening a window is much less. You open a program, get a window with focus somewhere. You can start doing your thing immediately without considering window management. Every new window on the stack doesn't affect the others you were using. Subsequently organizing the layout of your windows to be optimized for your current workload however takes more time. This appears to me to make stacking window managers to be more suited for people whose computer use is more dynamic or less organized - people can open and close stuff on a whim.

More organized vs. less cognitive load can be both a personal preference or a workload dependant thing. In case of tiling WMs it also doesn't help that they tend to have very low discoverability and depend almost exclusively on keyboard shortcuts versus GUI elements, only adding to their newbie-unfriendliness. I think that's also telling about the use cases tiling WMs are popular for.

I see a lot of people that respond to you by appealing to some objective superiority of tiling window managers, but there's advantages and disadvantages to both.

> With a tiling window manager, every window you open has cognitive overload. Which workspace, which location on screen, is it okay if other windows resize, or do I have it float even? Every tile you add affects the other tiles in some way - at the very least a new tile (e.g. on a new workspace) consumes space that cannot be taken by other tiles. The reward for this additional cognitive load is that every action is concious - reducing a lot of the mess that can pile while using a comptuer. This makes them incredibly suitable for people whose computer use is predictable, organized and focused on quick parallel tasks. I understand why so many developers use them.

This is not true, at least not in i3. When I don't want to think about how to position a new window I can simply switch to tabbed or stack mode, where all windows inside a determined workspace will be "maximized", and I can switch to them by changing the tab (mostly like browsers and web pages). This isn't really much different to Alt-Tabbing, except that I find more efficient (of course, this last part is IMO).

It is true that on my current setup almost all workspaces have a specific usage, however I have a "trash" workspace specially created to put those things that does not make sense in any other workspace. This workspace generally runs in tabbed mode thanks to above.

> With a tiling window manager, every window you open has cognitive overload.

I've been using a tiling window manager (XMonad) for ~3 years now; this is definitely true for the first few days (or weeks/months) until you adjust. After that, there is less overhead as you become more familiar with how the window manager behaves.

But this is also true for stacking window managers.

> With a stacking window manager, the cognitive overhead of opening a window is much less.

I think the level of cognitive overhead involved comes down (in the case of tiling window mangers) to what window rules you have, what kind of bindings you have set up to navigate between windows, etc., and how comfortable you are with those rules and bindings; the less comfortable you are, the greater your cognitive load will be when trying to use them.

Again, this is also true for stacking window managers.

> Every new window on the stack doesn't affect the others you were using.

Unless it "steals" focus and blocks the window you were working on. But that's expected behavior in a stacking window manager, so it's not much of a disruption. And it's one a user will certainly be aware of when opening a new window.

And this can (and usually is) be true for tiling window managers as well.

> I see a lot of people that respond to you by appealing to some objective superiority of tiling window managers, but there's advantages and disadvantages to both.

Right. There are pros and cons to each and really, but most people just don't really care that much about what window manager they're using.

I do want to add that one of the biggest pros to a tiling window manager (and a few stacking WMs) is the amount of customization available; if you're using a WM like XMonad, i3, Sway, bspwm, herbstluftwm, etc., you can configure the WM to do basically anything (including behave like a stacking WM), while many stacking WMs are fairly limited in how much customization they can have.

But that makes the assumption you're interested in customizing, which not everyone is.

What I've been thinking is, why can't we have both? Imagine just two changes: a workspace's mode can be switched between tiling and floating, and additional workspaces are just windows. Now you can have any combination of floating and tiling that makes sense to you. You could, for instance, set your desktop to tiling mode and open two new workspace windows, the switch one to floating and run your GUI desktop stuff on the left side of the screen and several terminal tiles on the right.

This is the mode how I run Windows with WinSplit Revolution.

It doesn't change anything by default, but with alt-keypad I move certain windows to the upper left, upper right, bottom right, bottom left and so most open applications have their exact position, but everything else is floating.

I guess I'd use a full blown tiling WM if there was one, though, I don't really see any downsides once you configured it to your preference. So bad for new machines, good for regular use. (for me).

I think AwesomeWM allows that

I used awesome for a few years and while I think you're correct to say it does try to do both, in reality it ends up doing neither very well, IMO.

I get the opposite feeling. I feel like I'm in a trash dump when I use an overlapping window manager, and have to sort through the pile to find what I need. Tiling window managers perhaps make more sense for devs or other professionals who use several of the same applications every day and know what window and pane they will be on, and can switch quickly with keystrokes. That being said, once you internalize the model and key bindings, it's not restrictive at all and can work for the general case as well, and actually feels freeing rather than boxed in. And there is always the option to break a window out from tiles and overlap if you so wish.

What I usually want to do is just click and resize to something that "feels right" for each of my programs, and then flip between them as needed, then resize again later as I workflow progresses. Often I want some important text from one to be visible while I'm working in another (thus overlapping), so I jockey them around until it feels right, and go.

Tiling manager remove the need to resize. You use virtual desktop to switch between apps. If you need several apps side by side, tiling comes in place.

Tiling manager are often keyboard driven. If you use mainly your mouse/touchpad, it can be quite painful.

My feeling in floating window manager is that i soon as a window is hiding another, the one on top should just use all the available space, there's no point in seeing the one bellow. If you need information from 2 windows, it's much better side by side.

You might like the minimal tiling present in CWM (and likely other window managers, too). Basically, it is a regular old overlapping window manager but with keyboard commands as the primary control. If you have two or more windows you want to see on the screen at the same time tiling style, simply press the hotkey for either vertical or horizontal (Super+v for me). When I want to open a terminal window on top of a browser, pressing Super+enter puts it at the mouse cursor, scaled to the application’s default size.

For many terminal windows or one Emacs and one terminal, I might use tiling, but I can still overlap whenever it makes more sense. Plus, the consistent keyboard controls (I use a config pretty similar to https://www.c0ffee.net/blog/openbsd-on-a-laptop/#cwm) make me more efficient than a mouse interface.

Funny enough that's precisely my use case for tiling window managers. I'm often referencing one app to do work in another, and often one is just a bit too overlapped to see what I need, and I also need to lay windows out every time I do this work. With a tiling window manager, you can be sure that your app is not obfuscated, and there are various adjustments you can make to pane size and layout to get the right amount of screen space and positioning for apps. You can also save and restore layouts for particulars workflows.

I think the key difference here is the use of multiple virtual desktops in the tiling case. In my setup I typically have one vdesk for browsing, one for docs, one split between code/terminal. I may have others open for chat and music, and I can switch between any with one key combination, or resize with another. I love that when I open a new window it will maximize the screen real estate of wherever I put it.

Granted, I've been using i3 for more than half a decade, but I do agree with sibling posters on how 'freeing' the experience is once you get used to it. There is no doubt a learning curve; but in my experience it's on the order of days.

I use tools that let me jump to an open window by typing its name. Spotlight on MacOS, equivalent tools exist on Windows.

I don't care how things are arranged spacially if they are just a few quick letters away.

Ideally you'd be using multiple workspaces, and only have 1-4 windows on a single given one.

I like to organize my workspaces by a category of activity. I'll have one per project I'm currently working on, then one for communications (Email, Chat, etc) one for web-browsing, etc.

I try to keep the non-project ones consistent, so at any point I can, say, tap Meta-9 and instantly be brought a web browser on Workspace 9.

I mostly switch between a full screen terminal and a full screen browser. The benefit of tiling is that when the browser opens a popup, or I need to look at my terminal next to my browser it's super simple. You just shift the browser to the workspace with the terminal with a keypress. Then when you are done you shift it back.

Tiling is simply more efficient it drives you into a keyboard based workflow that makes the most efficient use of your screen real estate. You can still have floating windows in Sway, or tabs for that matter as well when you need them.

I suppose it's a matter of personality, then. That kind of workflow would drive me nuts.

But there's nothing different between your workflow and my workflow except you have to manually resize your windows on each workspace. Sway just does it for me and I can save the layouts.

> But there's nothing different between your workflow and my workflow except for the fact that our workflows are completely different.

Is it so hard to accept that there are different ways of using a computer and your perceived increase in efficiency might not pan out for everyone? :)

I don't accept that it isn't immediately obvious that tiling is more efficient than dragging to maximize real estate on a single workspace. If someone wants to use an inferior method because that's what they are comfortable with that is fine. Just don't assert that the two are equal without some reasoned argument.

Some windows are more efficient to use when they are a certain size or shape, such as where text is made to wrap at a comfortable width for reading. Efficiency of screen pixel usage isn't the only factor affecting efficiency of the interface in practice.

You can resize, tab, and float windows in sway. You can also scale individual applications. There's nothing you can do with dragging that you can't do in sway faster with less presses and with saner defaults.

You can scale applications independently when they are floating. But that will affect the layout of other tiled windows on the same screen, as they will tile behind the floating window. So you would need to float everything. In that case it is not being a tiling WM

I wasn't actually the original poster :)

The thing is, we have studies that prove that mouse + keyboard interaction is faster than keyboard-only https://www.asktog.com/TOI/toi06KeyboardVMouse1.html - until you become super proficient with all the relevant the keyboard shortcuts.

So it's not about "inferior"/"superior", it's about your use case.

Another thing to consider:

Even if there were a significant difference one way or the other, it's rare to spend more than 1% of your time arranging your workspace and switching views & windows, so any optimization of that 1% will have minimal effect on your productivity.

> The thing is, we have studies that prove

"Originally published in the AppleDirect, August, 1989"

So? Do you think that users have changed so much in these past 30 years? If anything, things have regressed in some regards, as kids these days start directly with touch interface, single user, single application/windows OSes, basically, smartphone OSes (I know that technically mobile OSes are not like that anymore, but that's how they're used 99,999% of the time).

Funny, I feel the same way about overlapping WMs. I think it comes down to how different people perceive space. Vive la difference!

You know that there's multiple virtual desktops, right?

... which doesn't solve the underlying problem. I don't want to switch to a different desktop, such that my text editor window disappears and my terminal appears. I want them both on the same screen, along with my browser which points to a SO answer, in big sizes. In a normal window manager, this is accomplished by making them all big and then either overlapping them in ways that makes it easy to click on them, or just alt-tab. In a tiling window manager, this doesn't seem possible as far as I've been able to tell.

So you want to go to an app. You just hope that some corner is peeking out of that huge unsorted pile of windows. You call that good UX or is it just a formed habit?

Precisely. I arrange them such that I can click in the corner I need to get what I want, and then alt-tab between them, or click for the specific one I want if I need to jump. Then I get big real estate, and easy access. I call that good UX because it works well for me.

You can enable tabbed mode for a split (or the entire screen) in i3/Sway with a single keypress-- it produces the same effect as what you describe, just automated.

edit: You can also use something like rofi to fuzzy search for windows and quickly jump to them, instead of alt+tabbing.

I don't think it's "easy access" nor "big real estate". But to each their own.

There are more ways than clicking a window to give it focus. Alt-Tab being the obvious once but also Win+{Nr} on Windows if I recall correctly.

Alt-tab works fine as "go back to last window", but that could just be a specific shortcut.

Using it to cycle through a pile of apps shown as icons or screenshots is terrible UX, IMHO.

I love tiling window managers. I spent way too much time in non-tiling managers optimizing window sizes and so on - basically my 80% usage is to split the screen in half, one app on the left, one on the right.

This is basically what I do at work on a Windows system, using the drag-to-left and drag-to-right.

For your example case, I'd split the screen with my editor taking up the left half, the split the right half into upper and lower, with SO answer in top half and terminal in bottom half. That's the quick and easy split, I could also fiddle with resizing (using arrow keys) the split a bit left/right or up/down.

This gets the info all visible at the same time, without clicking or tabbing to bring one window at a time to the top.

The usage I described above is what I do on my work machine using i3. I'm gonna try out sway now that it's 1.0!

I can't tell if you were aware by your comments, but in a tiling window manager you don't have to have everything visible all the time. They all have modes that hide windows behind others and allow you to swap between them the same as alt-tab. Many of them also have the option to 'float' the windows as a normal non tiling window manager would.

But then I fail to see what benefit it's giving me to be a tiling window manager?

That's simply an option, because the devs recognize that not everything adheres well to a tiling workflow. It's not intended to be the primary mode of operation. I personally almost never use floating mode.

Better default window behavior. I do not need to use a mouse with i3.

That's another point of difference, then. I love the mouse! And the default window behavior is appealing to me.

> In a tiling window manager, this doesn't seem possible as far as I've been able to tell.

It's not and it completely defeats the purpose of it. Sounds like you shouldn't use a tiling window manager.

The one place where I prefer stacking window managers is that I can have a transparent terminal/editor floating over the docs. Stacking things on top of each other while being able to see all of them is pretty nice.

> I can have a transparent terminal/editor floating over the docs

Easily done even with a tiling WM. E.g. I have a keybinding which allows me switching between tiling and floating of a window.

a cool feature youll probably like is in tilling vms you can use stacked/tabbed views to get those benefits

Tried that, but then I can't have an important part of a SO answer visible while I work in my editor to play around with stuff, while watching the terminal churn away some stuff in a partially obscured window such that I can see when it finishes.

You can float windows in i3 if you wanted to do this. I also have a hotkey to make my window 80% transparent so I can reference docs while keeping my editor fullscreen.

But yeah, if that's your big use case, seems like you just prefer non-tiling workflows.

tiling wm is great when you have big screen with good resolution

I always code with two slightly overlapping windows, and quickly switch between them using alt-tab. I can't imagine a tiling WM to be an improvement, to be honest, because it will make my coding windows smaller. I'm curious about the other features, though.

It does require having the multiple desktop paradigm deeply embedded in your fingertips. I always used full screen windows and multiple desktops so switching to a tiling WM was natural and allowed me to use my available screen space much more efficiently.

> Sway is absolutely incredible, it puts macOS, built by Apple's army of engineers and dump trucks of money to shame in its simplicity, stability, and efficiency.

They have different design goals. A tiling window manager is not a good default for everyone or every device type or screen size.

> A tiling window manager is not a good default for everyone or every device type or screen size.

I disagree on the point of screen size. On small screens I want every last pixel to count, and tiling and tabs work well for that. On large screens, I find tiling to be a more efficient option than dragging windows around.

That said, available input devices absolutely do make a difference, and that does correlate with screen size.

I could equally state that overlapping windows are not a good default for everyone or every device type. It's also worth pointing out that Apples newest UI, iOS, uses a tiling window manager.

Tiling window managers were tried early on, when PCs were introduced. They failed because normal people didn't like them.

Also, mobile OSes won't really ever have more than 2-3 windows shown at the same time, since the screens are so small. That will probably change once we have folding devices or AR/VR. But super limited "tiling" window managers do make sense with such limited UIs.

> They failed because normal people didn't like them.

Do you have proof of this? I suspect they failed because everyone decided to copy the Mac which was trying to emulate a real desktop with bits of paper on it.

> Super limited "tiling" window managers do make sense with such limited UIs.

They also often make sense for complicated UIs, e.g. IDEs such as Visual Studio and browsers such as Firefox, which have docked sidebars and tabs in preference to overlapping windows.

From https://en.wikipedia.org/wiki/Tiling_window_manager#Microsof...:

> The first version (Windows 1.0) featured a tiling window manager, partly because of litigation by Apple claiming ownership of the overlapping window desktop metaphor. But due to complaints, the next version (Windows 2.0) followed the desktop metaphor. All later versions of the operating system stuck to this approach as the default behaviour.

The litigation from Apple came after Microsoft copied the Mac with overlapping windows (as shown elsewhere in Wikipedia), so this article does not appear correct. Windows 1.0 had tiling windows because IIRC, its lead developer, a former Xerox Parc employee, insisted on a tiling window metaphor. Many other multitasking systems of the period were also based on tiling windows. Overlapping windows certainly became the fashion, thanks to Apple, but you haven't proved they failed because users didn't like them.

Tabs are a completely different metaphor to tiles.

They are an example of managing content without overlapping windows. i3, Sway and Xmonad all have good support for them.

True, but Openbox, Fluxbox, etc. support that, without being tiling window managers. Heck, even Windows 10 now has support for window tabs :D

Why should you (upgrade from i3/X)? What's the benefit?

Based on Wayland, meaning it's more modern. Here's what I've heard for benefits:

* Better multi-window and gesture support

* Gaps built in (currently, i3 is separate from i3-gaps, which is what most actually use). No separate forks to deal with, meaning easier maintenance.

* Wallpaper

* Faster

However, there are some big downsides. Wayland is still pretty beta, and not everything works well with it. This is not the fault of Wayland/sway per se, more the fact that everything is based around X11. Biggest thing for me is no copy/paste. Hidpi scaling is supposed to be a nightmare. Also, sway doesn't support Nvidia, and the author has publicly stated he refuses to support any Nvidia drivers (or at least until they're open-source).

I think you have it backwards with regard to HiDPI. DPI scaling in X is a nightmare. Sway claims to have the best HiDPI support of any desktop environment due to modern Wayland support. Getting X applications to scale with Wayland HiDPI will require workarounds because of the lack of DPI support in X.

> lack of DPI support in X

Even back in 2002 RandR supported obtaining screen size information in pixels and millimeters which can be used to calculate DPI... or actually, what people really want when mentioning "DPI support", an automatic way to scale the windows.

How the window scales is up to the toolkit to implement and is independent of window system and really, even independent of screen size/DPI since some users (e.g. due to sight problems) may want their windows to be scaled even on "low DPI" monitors (the whole DPI thing is just a default setting).

The main problem i see with X here isn't some technical inability, but a way for window managers to notify other clients that they should change their scaling (since the window manager is in a better position to know what scaling each window should use - especially when you want different scaling per monitor, which again can be independent of DPI since you may want a large scale for a wall mounted TV even if the DPI would be very low) - essentially something like a new _NET_WM_SCALE window property for EWMH clients and window managers.

In what sense does Wayland have better hidpi support than X?

Mixed DPI works for one. I'm using GNOME on wayland currently but that's really the only thing I've noticed that's better than X.

As mentioned in the OP, we now have wl-clipboard. :)

Furthermore, it's proprietary NVIDIA drivers Wayland isn't supporting, not NVIDIA itself. AFAIK nouveau works just fine. There are many great reasons for not supporting the proprietary NVIDIA drivers.

Noveau sucks, unfortunately. Performance is bad, and you can't load custom firmware like you could with older cards (signing problems). I respect the effort the developers have made; any other driver would be of excellent quality had this much effort been put into it. But Nvidia keeps on being a jerk about it. Easy to say don't support Nvidia, but at the present moment they have the fastest consumer cards on the market.

> at the present moment they have the fastest consumer cards on the market

This is becoming a lot less true. Unless you really want to run the latest games (in which case you'll be on Windows anyway), nvidia is on equal footing as AMD nowadays on Linux. AMD has made a lot of efforts and progress the past few years, and them playing nicely with the OSS community makes a big difference.

My next computer will have an AMD card&processor.

I run Wayland on my hypervisor OS and pass through my NVIDIA GPU to a VM with proprietary drivers in order to take full advantage of my card.

Do you know of any blog posts that would discuss a setup like this? Sounds pretty interesting.

Alex Williamson has an excellent VFIO blog that discusses the basics of GPU passthru. [0]

I think the instructions are meant for Debian however and there are some differences in the commands you need to run. If you use a Redhat distro and/or you want some further optimization tips I'm also happy to work you through the process.

https://vfio.blogspot.com/2015/05/vfio-gpu-how-to-series-par...

Awesome, thank you!

Sway was the reason I chose a Radeon card to my office workstation. Been running Sway from Github since beginning of this year and it is very stable, fast and great experience, supporting my old i3 configuration as-is.

Now I know my next home gaming computer will also have a Radeon card.

P.S. Put a monthly donation for Sway. Really appreciate the work!

I never understood the allure with i3-gaps. Without gaps wallpaper is only relevant for empty workspaces which I dont really use. The no copy and paste is a nonstarter for me.

No copy-paste is a non-starter for me as well, and probably the largest single reason why I still am on i3-gaps. I still like wallpaper mostly because it's nice to add some light transparency to console windows. I3-gaps, for me, was initially because I had a very, very old computer only. This thing had been retired by a corporation three years prior; ancient i3 CPU. With i3-gaps, it ran nice and snappy; on Windows 10, it largely choked. I've appreciated it ever since.

One of the biggest draws is keyboard-friendliness: you don't have to use a mouse to navigate, ever. Nice for those of us who already prefer programs like vim that allow keyboard usage, and can increase productivity. It's also good for multi-window setups (having more than one open), though some people who use tons of windows use bspwm (represents windows with a binary tree).

Vanilla i3 is also keyboard / low resource friendly; for these things i3-gaps is no different than vanilla i3. Having a wallpaper or transparent terminals never did anything for me but maybe become distracting.

Not having to download i3-gaps or feh in exchange for copy/paste and hdpi scaling not working seems like a questionable tradeoff. I'm not sure what 'better multi-window' support means? I've never had issues with i3 windows. Speed has also been non-noticeable.

So it seems like the feature list is 'better gesture support' in exchange for a lot of downsides?

I am not trying to be dismissive, but I see so many people evangelizing it over i3, I feel like I have to be missing something. I can't believe there's a single person who finds those tradeoffs acceptable outside of those who want it to be good eventually and are working towards that.

Again, this is what I've heard; never used it myself (mostly because of the issues you mentioned; I'm unwilling to give up copy-paste and hidpi). With that said, X11 is quite arcane. I3 is still good, or at least for me. Not going to try to proselytize here. I mostly switched because I had to use a completely ancient computer for a while, and I needed to use a tiling wm to get decent responsiveness.

HiDPI scaling is actually much better on Wayland than on X11. On X, if the app doesn't support HiDPI, the compositor cannot scale the window because it cannot scale input. On wayland, this is possible. Wayland also supports mixed-DPI multimonitor support. On weston, window dragging between mixed DPI monitors is seamless.

I've always wondered what's the use of gaps.

Twofold: for many, it just looks nice. But for me, having a little margin in between things helps me visually distinguish and makes it easier to look quickly at different windows. This is actually helped with a wallpaper, because it further accentuates the borders of windows. I guess a magenta outline on every window would technically serve the same purpose (the sort of thing you might see on a high-contrast UI), but it's a little less gaudy and a little more visually appealing this way.

I use a subtle (~20%) transparency on inactive windows for that (though with XMonad, not Sway). It helps distinguish windows, lets me see which window is active without having to look for the border, and wastes less space compared to either borders or gaps.

The desktop number 5 with cmus, alsamixer and a terminal music visualizer with gaps and transparency for eye candy.

Not everything needs to have a real use case!

What does better gesture support mean in this case? libinput works with X too and it can be used to enable some kinds of touchpad gestures there

Xorg usually ships with a different input driver, IIRC. Libinput is the default in Wayland and works seamlessly; Xorg actually needs a wrapper (xf86-input-libinput). Haven't personally used it, just what I've heard.

Because you can support Wayland and make Sway better. X11 is a super crusty, hacky mess that is being phased out. Sway also incorporates some nice features (gaps) that were previously not in the base version.

I really don't get it why people like gaps so much, yes it's look pretty, but in my opinion it is as much waste of space as window drop shadow. I would like to have as much screen real estate as possible to be used to show information, that's the whole point to use tiling wm in the first place for me.

A 5px gap between windows is a drop in the bucket, but adds a lot in terms of aesthetics.

I tried to switch from i3 a few months ago and spent several hours trying to get it to work with my Nvidia card before finding out that the project doesn't support Nvidia for ideological reasons.

I'm not well versed in the details of the conflict, and of course any open source project is not compelled to support anything they don't want to. That being said it would have been nice to know that they have no plans to support a large percentage (majority?) of the consumer, dedicated GPU market.

If/when I am able to get a new AMD card I will certainly try it then.

It merits re-empathising that it's Nvidia that does not support sway - not the other way around.

That's fair, but if your stance is that users should be educated to vote with their wallets, then I wonder if you've considered adding a disclaimer on the site saying "Nvidia users don't waste your time, here's why"?

The makers of the world's best and most popular graphics cards will never compromise their goals to accommodate an i3 clone. I think it's fair that the smaller entity support the larger one.

Our policy bubbles down to wlroots, which is the basis of a huge variety of Wayland compositors.

However, we only need to support Nvidia if our goals include "adoption by Nvidia users". But they don't, so we aren't going to.

This isn't an i3 vs. nvidia situation, this is linux kms/drm progress vs. nvidia.

nvidia is simply being uncooperative.

Those kind of corporate bullying tactics are not particularly welcome in open-source communities.

How is this on Nvidia? I'm on KDE neon and so no Wayland due to Nvidia's lack of APIs