Hacker News Reader

Comment by stormbrew | Hacker News Reader

Everything you're describing in this reply just exhausts the hell out of my brain. You're describing an incredibly large amount of cognitive load you hold in your head all the time about where things are and what they mean and how much of them are visible.

Tiling WM are largely about just not thinking that much about windows at all. They're visible or they're not, they occupy precise areas on your screen, and moving them around is easy and simple and they don't scatter to the 4 winds when you plug your laptop into an external monitor and then unplug it again.

Every time I have to go use a computer with overlapping windows I suddenly find myself thinking about where things are about 10 times as much and it's frustrating and tiring.

View on HN

What you’re describing there is a little like the difference between those who like to have tidy desks vs those who don’t.

You’ll often find that some who work in organised chaos still know exactly where everything is.

Is their organisation worse than those who keep things tidy? From an outsiders perspective it might seem that way. But if they’re the only ones using that workspace then it doesn’t really matter.

For some people, remembering where stuff is creates no extra cognitive overhead. For others, it’s like trying to access their computer while it’s covered in bubble wrap.

> You’ll often find that some who work in organised chaos still know exactly where everything is.

I don't know where things "go", but I know where I last put them!

> if they’re the only ones using that workspace then it doesn’t really matter

Yup, both methods have the exact same weakness - someone else moving something I need.

Of course, as you imply, everything should have it's place in a shared workspace.

I never have any idea at all where I last put things. I just remember 5 to 30 minutes ago I had something... and that's all I know.

I've pretty much always relied on full searches of my entire surroundings, or very long term plans with labels and everything, since remembering a new location takes active effort or weeks of time, I don't just see things and passively remember where they were.

The desktop completely rearranging itself randomly doesn't even bother me, as long as I can quickly scan the task bar, since I basically never use Windows any way other than maximized.

I don't see the cognitive load you are talking about.

It's just windows. You put roughly where you want them when you need them to have what you want visible. They stack. If one is out of your sight and you need it, you use the list or expose to find it and move it where it will be useful. It's like manipulating documents in the physical world except it's always easy to find a window in your stack of windows. It's a very natural flow.

Tiling is the solution imposing an unnatural cognitive burden on the user. With tiling you need to constantly think about exactly what you want to see, where you want your windows and you constantly need to fiddle with the arrangements and spend time preemptively positioning your windows. It gets very annoying to me as soon as I want more than one full screen window.

I’ve been the same. I came from MacOS into Linux in the last couple of years, but gave Pop_os! a try early on with its tiling wm.

I found myself constantly planning layouts and positioning things. Nothing ever ended up feeling right, and it was exhausting.

“I’ll have my browser window take the full height and half the width over here. Then I’ll put my code editor over on the right, half high, then my Terminal below it.

No wait, that’s not enough height for the code editor, so let’s shrink the Terminal down… except now the terminal’s uncomfortably short.

Well, maybe the browser can be split with the Terminal and the code editor can be full height.

Except now the code editor feels too narrow and I don’t really need it this tall, it would be really nice if I could have the browser window full height instead.”

And then I’m back where I started.

And if I finally get used to a layout and whatever shortcut keys I have for jumping around, then !!! A wild Slack message appears!

Now I need to open a PDF viewer along with Slack along with my code editor but don’t need my browser but maybe need my Terminal, and I’m back to the layout drawing board for the next task.

With overlapping windows, every window can be the exact size and shape it needs to be.

I’m open to the likelihood that I’m just doing it wrong. I would also bet this is less of an issue if I were using multiple large, unscaled 4K monitors.

So I’ll try it again someday with a different setup, but for now I’m mostly happy with my 1440p monitor(s), and overlapping windows.

If your future 4K monitors are the exact same physical size having higher resolution wont make you want to have more tinier windows.

The most useful arrangement for windows to be tiled into is 1-2 windows on a horizontal monitor with more windows accommodated by adding more monitors to your workspace instead of splitting a window 7 ways.

Psst. Try PaperWM. It allows quick resizing and tiling.

I can't speak for everyone and I'm a multiscreen type, but for me I have things on specific workspaces.

Workspace 1-4 Primary monitor (center, ultrawide, landscape)

Workspace 5-7 Second Monitor (left, portrait)

Workspace 8-10 Third Monitor (right, landscape)

1. Browser and terminals/vim for current project (ultrawide monitor = 4:3ish split)

2. As above if I need to switch to another project

3. As above if I need to switch to another project

4. Spare/"Personal" slacking off things like HN

5. Comms Slack/IRC etc...

6. Spare/Displaying things that work will in portrait like reading PDFs

7. Spare/Monitoring specific things to do with what I'm working on

8. Spare/Background noise videos

9. Music player

10. Email/Calendar

For me I usually sit with 5, 1, 10 visible across the 3 monitors 95% of the time which means I can see everything I immediately need. However if on a single monitor device the workspaces have the same use (mostly, 1 might get split into 1 = Terminals, 2 = Browser because it's a bit cramped on 16:9).

This means If I need my emails I hit Mod+0 which then focusses my email/calendar whether it's visible on my 3rd monitor or hidden on a single monitor setup. Or Mod+5 and I get Slack.

Once it's in your muscle memory you don't even think about it. The way I see it is it's like vim, once you have the shortcuts in your muscle memory you don't even think about where things are, they just appear when you need them.

Yeah I do the same with virtual desktops, except I right now have only one screen. Because I hate the way KDE switches virtual desktops for all screens together, I would really want each screen to have its own separate virtual desktop. I have 9 of them and mapped my numpad to the virtual desktop because I don't like combo keys very much. And I don't normally use the numpad for anything anyway :P

How did you manage that? Or don't you use KDE but something like i3? (Especially because you call them workspaces).

Ah yes, forgot that bit. You're correct I use i3.

Keep in mind that to some degree workflows are inevitably a matter of taste and familiarity. That is we adjust over time to a certain way of working changing how we work to be in line with out our tools work and enjoy things whose workings are in line with our own mental models.

I think you may have misunderstood how one works with a tiling wm rather than fiddling with the arrangement constantly one hardly does at all. Certain applications can be automatically routed to certain monitors and in certain arrangements. 99% of the time the primary window management operation one does is simply to switch what a given monitor is displaying or change focus between applications with none of the sizing arranging exposing or alt tabbing through lists of windows. The most useful arrangement of windows on a per monitor workspace is either a single large window or 2-3 aligned horizontally or vertically based on the orientation of the monitor and the creation of the arrangement is by no other means than simply opening an application and having it pop into the arrangement.

This is increasingly evident and useful with multiple monitors and increasing number of windows. There is probably no meaningful difference in utility between any window management method that can be discerned with 3 windows on a single monitor but at some point between 3 windows and say 15 one really needs at least virtual desktops to break up the windows and switching windows is a 2-3 level operation switching virtual desktop then focusing the correct monitor and then expose.

Much worse if it it would useful at the moment to have an application from Desktop 2 Monitor 2 on Desktop 1 Monitor 1 and then back again in its original location. Especially if you need it to share space with an existing window because now one most resize both windows in addition to rearranging.

The cognitive overhead is in remembering which desktop one has which window however this is trivially mitigated by always using the same desktop for the same task. This ultimately microscopic amount of overhead is paid for by not constantly paying for windows arrangement amnesia by using overview and expose to find the window you were using 30 seconds ago.

I pretty much never use virtual desktops. I never got the point. I want my windows to always be available not be hidden somewhere else. They stack perfectly fine and now that starting things is nearly instant I just close the applications I’m not currently using (I still have between ten and twenty windows open but that’s perfectly manageable with one desktop). I can get wanting windows side by side however but most descent floating windows managers have that nowadays.

Well, I guess it’s another thing I will never understand with people having trouble managing their hundreds of open tabs. It’s fascinating to me how people actually have very different way to use their tools and how it much shape the feelings they have towards them.

You don't need to constantly think about exactly what you need to see. Nor do I need to fiddle with arrangements. There's less fiddling. With the WM I use every window is available in every frame and I can just cycle through them. I don't need to remember what a window is called or distinguish between 5 nearly identical thumbnails because I'm cycling through the windows at full size instead of looking at list or expose. I don't need to think about what monitor some window is on or what virtual desktop and there's never any visible clutter or noise because anything that's not something I want to look at is just not visible.

Right, for me the default in a tiling WM has always been one fullscreen pane, which I can split on those occasional situations where I want to see more than one window at a time. Exactly the same as how I use screen or vim (which I use splits in far more often than my WM). If I'm not using a window right now I don't need or want to see it.

So if everything is visible or not, then isn't the cognitive load now on keeping track of all the windows that aren't visible? How do you handle "secondary" windows, so like something you don't need right now, but will in a bit, or you're checking in on it or using it from time to time.

So then overlapping windows serves the requirements of:

- constant visual reminder of which secondary windows you have

- but occupying minimal window space so as to be effectively ignorable if you're concentrating on your primary windows

- that might be able to show some visual indication in the corner of your eye when focus is required

- that can be easily switched to be the primary window when such indication happens, and then back again to the secondary state

KDE has two solutions to these problems: Present windows and desktop grid view. Present windows also has a search feature, so you press a key combination, present windows open. Two or three letters for the title of the window, then enter.

If you want to fast organization, showing all windows and desktop is also achievable. Open the view, drag and drop windows or desktops.

For me, both views have their active corners and keyboard shortcuts. People give me sideways looks when I use both of them.

Yes indeed, another thing I really like about KDE.

macOS used to have the grid view (with virtual desktops in a grid), along with Expose, basically like KDE has it now. But at some point (a long time ago) they decided I shouldn't want a grid view anymore, reduced virtual desktops to a horizontal line only and combined it with Expose, and called the whole thing "mission control". Since then my use of it reduced dramatically because it didn't fit my workflow anymore.

It was one of the many things they broke over time which built up and eventually caused me to leave macOS. KDE still gives you choices. Apple will just take away what you've been using and often gives something worse back in return.

For one, the grid view is really great if you're monitoring multiple things. I have 9 virtual desktops and I often use it to "zoom out" so I can see multiple screens at once. This is especially useful because I run my 4K screen at 200% scaling so even zoomed out it's still fairly readable (about 60% of normal pixels)

Every time I try a tiling WM again the cognitive load of memorizing new key bindings and trying to get various apps to cooperate with this exotic scheme pushes me away again.

Also many apps have their own built in tiling or separation. I use emacs, I use CLion, irsii, etc and I use a web browser. None of them have agreement on how tabs/documents/windows can be broken up.

I like the idea. The practice of any of them has been less than desirable.

> Every time I try a tiling WM again the cognitive load of memorizing new key bindings and trying to get various apps to cooperate with this exotic scheme pushes me away again.

Some tiling WMs are really into the whole "a keyboard should be all you ever use" thing and I don't personally like that. One of the reasons I like i3wm in particular is it's not extremely mouse-hostile and the number of key bindings I have to keep track of is pretty low and most of them make intuitive sense (mostly variants of <modifier-set>-<direction>.

I like the idea of meaningful sequences as bindings. o->t open terminal w->f workspace f r->f relocate focused window or windows to f y->f grab whatever is in f and put it in current.

Naturally you need a binding to start the sequence. Personally I like setting it up to produce a different binding if shift keys are pressed and released because this is much easier to press than a modifier key + key or a key which would require moving away from the home row.

Your comment and the comment it is a reply to remind me of the post-show part of the classic ATP episode “The Windows of Siracusa County” where John Siracusa explains his approach to windowing, to the absolute dismay of his cohosts because of the seemingly insane amount of complexity.

https://atp.fm/96 (at 1:29:30)

That's because you're first overlapping windows experience was Windows for sure.

With virtual desktops you just have 2 opened windows at most per workspace.

> With virtual desktops you just have 2 opened windows at most per workspace.

My first experiences with overlapped windows was actually probably Turbo C for DOS. :P Windows 3.0 and Macintosh System.. 6? after that. I've been through a lot of them.

I find nearly all overlapped windowing system's concepts of virtual desktops nearly unusable and incomprehensible, and I have used many. I'm a daily polyglot with operating systems, I use macs and linux for work, windows for play, and linux for everything else.

The only version of virtual desktops that's ever actually worked for me is i3's, because it's simple and easy to use.

Reply on news.ycombinator.com