Hacker News Reader

I don't understand how you can objectively qualify a tiling manager as "broken".

1st of all, tiling managers are a matter of personal preference, no wether it's "broken".

2nd of all, I've used i3 plenty, and the amount of power it gives me when I'm working is immense. I no longer need to alt+tab 7 times to get to the terminal for the monitoring servers, I always know they're on screen 4. I need my web browser ? I know it's in screen 2. Heck, I can pin apps, and even screens to specific monitors. When I'm working on an incident, my throughput is basically multiplied by 3 as a result of how streamlined my work environment is.

That is simply not possible with other environment where the mouse is the main device. Yes you have virtual desktops, but try having 15 open terminals without having them pinned in some form. Just a simple swipe to the corner and your GPU will heat-up from the 3D acceleration required to show all the windows :D

All-in-all, I genuinely believe that tiling wms are an amazing godsend, but for my use case. And I cannot emphasize this enough. It's a personal preference. Yes it's a bit of a pain to set up (although I literally used a two-part youtube video to set up i3, and never touched it again, not even need the Arch wiki), but if you don't have to tinker in your configuration are you really running linux :D

> I don't understand how you can objectively qualify a tiling manager as "broken". 1st of all, tiling managers are a matter of personal preference, no wether it's "broken".

People in IT love to declare their subjective experience as objective scientific fact. I don’t know if it’s because they’re taught that the industry is a science (albeit computer science) or if it’s just their inability to disable their analytical part of the brain but you see the same bullshit posts all the time from programming language debates, $SHELL preferences, desktop environments, operating systems, etc. IT engineers arguing other mundane preferences are as old as vi and emacs.

I am not talking about this as a preference, I think they are actually a fundamentally flawed design. This is my experience from using it for a long period of time and talking to hundreds of people about this and reading probably in the thousands of github issues and comments where the various bugs pop up. Is there some other better way you know of to identify issues like this? If so I would like to know. Otherwise it is somewhat specious to say that I don't know what I'm talking about. How long to I have to deal with bugs in this stuff before my opinions don't get handwaved away as bullshit?

1. There is no single design for “tiling window manager”. Some uses lists, trees, manual tiling… All can be inadequate. 2. The “tiling” part is no longer the only feature, most have floating modes, automation interfaces, … 3. Bugs happen everywhere.

For 1, all of them are inadequate in the same way, because tiling is just not adequate for most use cases. Most desktop apps are not designed for tiling, and the few that work well with it, also work just fine with another system.

For 2, if you have to use all those other features, I think that defeats the purpose of it. I can't count the number of configurations I've seen that float the majority of windows. What is the point? I ask people about this and they know it's bad but they just keep on going with it, it's like nobody wants to actually admit it's a problem. I'm sorry if I'm getting worked up about this but it's really frustrating to keep having this same type of conversation over and over again where nobody actually wants to fix any problems.

For 3, See above, there are specific bugs here that don't need to exist and could actually be fixed, quite easily I might add, but are purposefully being ignored because people expect it to be that way as part of the design. To me, that's pretty inefficient, and if it can be improved, I'm going to keep calling it out because I think people deserve the help. Maybe that's my character flaw, I don't know.

Maybe channel that energy into fixing those bugs yourself? Or writing your own WM.

That’s what I did.

I have already done that for quite a long time, that doesn't fix the problem, that just makes it so there is yet another half-assed and broken WM to add to the pile. Or should I say, graveyard, because so many of them have come and gone over the years...

Sorry to be blunt but that is another myth that I see going around in open source communities, that the way to fix anything and everything is to just sit down and write more code and try to close as many github issues as possible however you can. It's not. That only gets you so far and in a lot of cases it can make things worse.

You’re the one who said there are bugs to be fixed. I’m literally just reacting to your own statements. If you think it’s a myth then don’t introduce it as a cornerstone of your own rant to begin with.

Ok I see what you're saying. Sorry for the misunderstanding: the bugs here can't realistically be fixed. They keep popping up over and over again and they will for as long as the design stays flawed. It is the design that needs to be "bug fixed" if that makes sense. I am not talking about any myths in relation to that, although there are several others I could name after reading various github comments...

> Most desktop apps are not designed for tiling

I'd say this is the issue, not that of tiling managers themselves.

I didn’t say you don’t know what you’re talking about. I said you’re conflating opinion with fact.

I have not done that, in fact I intentionally used the phrase "in my opinion" in my original comment so I wouldn't get this type of response. But I guess I didn't use it enough times.

Opening statements like the following undo any charitable interpretation that you’re separating the two:

> I am not talking about this as a preference, I think they are actually a fundamentally flawed design.

Please do not undo charitable interpretations. I intentionally put the phrase "I think" in that comment so it was clear that it was my opinion. I don't know how to make it any clearer than that, it is a thought that I had. It is not some kind of universal truth. What else do you need from me?

Edit: If you'll notice I also used the phrase "my opinions" farther down in that post...

You seem a nice guy and this is clearly a passionate topic for you, I just don’t understand why you’d fail to consider that tiling WMs might work for other people despite their short comings for you personally.

Maybe those people have different workflows and needs to yourself. Maybe they enjoy the mental model of tiling so much that these showstopper issues you describe don’t even amount to minor annoyances for others. Or maybe you’re looking at this from too much of a purist perspective and missing the benefits of a tiling-first but ultimately still dynamic WM.

If there is one thing I’ve learned in my life, it’s that things which seem like insane choices to me are often quite reasonable for other people.

Well I said this earlier but this is not specifically about my personal experience with shortcomings, although that was an influence. I have heard many of the same issues coming from large numbers of other people, and I think there is a reluctance to really own up to it because of some kind of fascination with this "romantic" idea of how a keyboard interface is supposed to be. It's like, people get that idea in their head that's what they want and they can't let it go, and then it becomes a game of sunk cost fallacy. Of course that is not new or even all that uncommon in open source but I think it's irresponsible to not point it out when I personally see it.

"a tiling-first but ultimately still dynamic WM"

In my opinion, these are all bad hacks that are made worse by being tiling first. In some ways I would actually consider these worse than a pure tiling window manager, because they tend to not focus on any particular way of doing things and so it becomes a pastiche of concepts that are not really compatible.

What kind of bugs (or fundamental flaws) I'm wondering? I used i3 for over a year (before switching to Mac due to work) and it worked great. I was annoyed at confirmation dialogues taking up half a screen but that's the only issue I remember ever having.

I am telling you this as someone who had that personal preference for at least 10 years. They are broken and they don't work, the uses cases you use them for can be done better with anything else. I'll go through them for you, keep in mind I am not doing this to challenge you. I am doing this because I have had all exactly the same thoughts as you in the past, and my thoughts all changed when I decided to look at the problem in a different way. In my experience, all that staring at terminals all day messes with one's head, it makes it very hard to see other solutions that could exist outside a very terminal-centric viewpoint.

"I no longer need to alt+tab 7 times to get to the terminal for the monitoring servers, I always know they're on screen 4. need my web browser ? I know it's in screen 2"

Binding an application to a workspace or a key can be done with pretty much any window manager. But you don't want to do that because "screen 4" is abstract and doesn't actually mean anything. You are putting this through an extra level of abstraction for no real reason, it would be better and less confusing just to have some way to identify the correct app.

"Yes you have virtual desktops, but try having 15 open terminals without having them pinned in some form."

This is not actually a good feature of the desktop, this is all a workaround because of the other horribleness of various linux terminal programs that don't offer a good way to distinguish between windows. Also, you can upgrade your GPU you know? It is worth it in general if you want a faster and less laggy display.

"And I cannot emphasize this enough. It's a personal preference."

Respectfully, please stop emphasizing this, it doesn't add anything to the discussion. It is the equivalent of saying "well that's just your opinion". Of course it is our opinions and preferences, but we are here to discuss them. Right?

"if you don't have to tinker in your configuration are you really running linux"

This I think is part of the major problem. The Linux community seems to intentionally gravitate towards things that need more tinkering just for the sake of doing so, and actively resist efforts to reduce the amount of tinkering that needs to happen.

When tiling window managers don't work for you, that is totally fine, but I wouldn't say that they are broken and awful. They all have different trade offs, the same as all the Desktop Environments.

I tried a couple tiling window managers many years ago and none was usable in the way I thought how it should work. But instead of investing time to make it work the way I want it, I just switched to the next one and tried that. I think at that time I tried 5 or 6 window managers a day.

The one that worked for me at the end was one written by Sean Pringle called Musca. It used relative movement from each window and it made so much more sense than all others. Sadly he dropped it a bit later for another project but herbstluftwm is its successor and it worked for me. My config hasn't actually changed for the last 10 years.

There are so many great tools out there we can use for what we want to do. Choose the one you like and works for you. Thats why peoples desks or their work places all look differently. A smith woulnd't talk to a smelter about his setup being awful, because he does a different job.

I don't understand your analogy. If a smelter is using something in a bad or inefficient way and a smith is capable of explaining it to them, then that would be beneficial to do so. People do not have to do exactly the same job in the exact same way in order to discuss how a specific tool works.

Like you, I went through a period where I tried pretty much every tiling window manager I could find. And I can say with confidence: they are all broken and awful, in part because the very concept of tiling doesn't really work. The design is flawed. It's like you are making a comparison between every jigsaw on the shelf and picking out the best one but are missing that you could get a miter saw and that would work even better for your use case. Does that explain it?

Is musca the one where you divided the screen first, and later could fill specific applications into those areas? I found that pretty cool at the tkme!

It could probably work much better with wayland though, because programs often refused to remain in their little borders.

> tiling window managers are awful and broken

You comment may have been useful if you would have said how and why they are awful and broken. For now it is just a rant without any supporting arguments.

I'm sorry, the article does not mention anything specific really so there was nothing I could respond to, I will follow up to a post above though.

In what way?

I found Amethyst on macOS to be extremely useful in bridging the gap between how modern UIs/window management wants to work, and tiling window managers.

I'm curious what makes you see them as broken, and whether that's specific to the deeply integrated X-compatible ones?

In my opinion, all that stuff on MacOS (Amethyst, Rectangle, Spectacle, etc) is also broken in the same way. Apple's window manager has a built in fullscreen tiling mode that could actually be way better but is currently missing things. I think there are some paid window managers there but I haven't tried them, maybe those are better? I don't know.

You can read some of my other comments, but with MacOS it's the same kind of problems, part of it is that the apps don't play nice with it and they can't really be made to do so without breaking things.

What? How are they broken? I can't imagine going back to anything else ... I really never had a better interaction with my computer than with i3.

You used a "broken system" for 9 years ?

Yes, if you think it's stupid, just imagine how I feel...