A dedicated key for all window-manager things is what people that have thought about it do (I use the "windows" key). But keyboard manufacturers haven't thought about it, so sometimes reasonable things aren't possible. I don't know.
That's fine as far as it goes, but I don't think that gets you what this article is for, which is things like using the same binding context-dependently to navigate between emacs splits and regular window manager windows, context-dependently. Which is a fun bit of overengineering.
The problem some of us have is we get used to jumping between emacs windows/buffers using C-x o and C-x b and then without thinking about it try and use the same keys to jump between i3 windows and of course it doesn't work. Or vice versa, trying to use i3 shortcuts to switch emacs windows/buffers.
I keep trying it (coming from EXWM) but I get lots of lag, stutters, and poor fractional scaling. I'm not sure how much of that is "GTK under wayland", Emacs's PGTK build (known to have lag/rendering issues), AMD kernel drivers (?), or EWM itself; but it's not yet a replacement for EXWM in my experience.
"Writing a Wayland compositor from scratch is a staggering amount of work..."
I switched to Wayland and sway about 2 years ago and it hasn't been bad, but this is the thing that makes me most sad about Wayland. X has so many different window manager options, it's like, an embarrassment of wealth. Wayland has like, 3 "finished" ones (if sway even counts).
it is sad, X11 had a suite of tools like xrandr that worked regardless of your wm/compositor but now with Wayland these tools are compositor-specific (or have to agree to a standard).
Side note: I am really enjoying HN today with the set of stories with personal hacks like this i3-emacs integration, someone's desk setup, someone's writer-deck laptop install, the kinda hilarious but also hecka geeky thermal ttrpg thingamabob, and the 16 byte wake up demo. Fun geeky stuff that isn't AI,and I love AI, but it ain't everything.
The problem now is that so many articles with innocent titles are really about AI slop. The title might be “A Space Exploration Game” and the content is about some vibe-coded garbage.
In my current favorite alternative HN frontend https://hcker.news, you can filter (include and/or exclude) posts by words found in the titles. There's _a lot_ of other customization options as well.
Ooo this is nice. I may have to try to get this working with my personal setup using Emacs and Sway.
My long term vision is to make an Emacs implementation that is compatible only in philosophy. It would use Guile instead of Elisp, default to bindings that are more familiar to people coming from more modern systems, and would be built from the beginning with concurrency and graphics in mind. For now it remains a dream though.
Nope! I should have elaborated, but by "graphics in mind" I meant full support for graphical applications. I want it to be a Wayland compositor. It would either be used as a top level compositor like EXWM, or as a nested compositor, like how gamescope is used.
I want it to be as easy to make scripts to automate graphical applications as it is to automate textual ones in Emacs or shell scripts.
As I said in my other comment in this topic I actually would love to see an arch where the UI portions are split up with a background daemon holding buffers, lisp execution etc and then IPC to frontend pieces for window management and buffer editing.
So window management can be done by ... a window manager, but with intelligent interaction to the editor pieces so you don't lose all the awesome emacs stuff.
EDIT: I would say however that something like lem is probably more amenable to that refactoring/restructuring than GNU emacs, which is a single-threaded monolith.
I'm familiar. The difference is that this is targeting source compatibility with existing Elisp. I don't feel like that is worth it for most people who would be interested in what I want to build.
I did the same integration with an Erlang daemon. All relevant key presses are sent to it and based on the current focused application the daemon does different things. I built an Erlang library i3_IPC to listen for events and send commands to Sway.
In case you don't understand the problem here, an emacs instance can be split into multiple "windows" and there are emacs key bindings to create and destroy these windows, move the "focus" from one window to another, resize the windows, etc. For many of us, this was our introduction to a tiling window manager experience before we'd ever heard of tiling window managers.
When we switched to using a tiling window manager for all our windows, we ran into a muscle memory problem. We were used to jumping between emacs windows/buffers using C-x o and C-x b and then without thinking about it we'd try and use the same keys to jump between i3/sway windows and of course it doesn't work. Or vice versa, trying to use i3/sway shortcuts to switch emacs windows/buffers.
To try and solve this problem I've been using less emacs windows and more i3/sway windows, so I can just use i3/sway keybindings everywhere, but emacs puts up some resistance to that. I like this approach
Yep, I still find emacs' window management bindings more intuitive than any tiling window manager I've used. So much so that I thought naively that things like exwm would "solve" this problem for me. But in fact -- beyond being janky / single-threaded -- they don't really because you still end up with competing bindings.
I think the more elegant model would be for the window manager to own all the window management keybindings and then for there to be a background emacs-like process that owns the buffers (and the remaining keybindings) and can present them in standalone windows which the window manager presents. With some kind of IPC between the two for coordination, I guess.
I don't think GNU Emacs itself works well with this model tho. Its "server" mode is something else entirely.
I can relate, finding myself pressing the wrong keybindings when moving between i3 and emacs windows. The idea in the post is interesting but I am not sure how I’m going to solve this. I have been thinking about trying EXWM or switch to another WM. Tried KDE but it didn’t have support for multiple workspaces in the secondary screen. And after using i3wm I found it difficult to get back to other WMs.
I'd be surprised if someone hasn't already solved this. I'm not an Emacs user myself but try searching for something like "seamless <tiling_system> <tiling_app> navigation site:github.com". I've seen several scripts/plugins to setup seamless navigation between i3/tmux, tmux/Vim, i3/tmux/NeoVim, i3/Zellij etc.
Recently I hacked together a Bash script for seamless navigation between i3 windows/tabs and Kitty windows/tabs using i3-msg, xdotool, jq, kitty's remote control sockets. Now if I only figured out how to add Helix windows into the mix...
Sometimes thinking outside the box helps. I've been able to unify a lot of keybindings by using a context aware remapping tool such as Keymapper (https://github.com/houmain/keymapper.
I have to mainly drive Windows 11 for work reasons so I use komorebi for a tiling window manager (highly recommended btw). I bet the ideas in this post can be applied in that environment too, gotta try it now, great idea
PunchyHamster | 18 hours ago
royal__ | 17 hours ago
dima55 | 17 hours ago
sudahtigabulan | 11 hours ago
rileymat2 | 17 hours ago
noosphr | 17 hours ago
https://xcancel.com/octonion/status/1341113219142828039
topaz0 | 14 hours ago
krupan | 6 hours ago
skulk | 17 hours ago
https://codeberg.org/ezemtsov/ewm
stebalien | 14 hours ago
krupan | 5 hours ago
I switched to Wayland and sway about 2 years ago and it hasn't been bad, but this is the thing that makes me most sad about Wayland. X has so many different window manager options, it's like, an embarrassment of wealth. Wayland has like, 3 "finished" ones (if sway even counts).
skulk | 3 hours ago
SubiculumCode | 16 hours ago
diminish | 10 hours ago
torh | 7 hours ago
Two-part desk setup : https://news.ycombinator.com/item?id=48214311
Writerdeck: https://news.ycombinator.com/item?id=48250144
Wake-up demo: https://news.ycombinator.com/item?id=48253060
noir_lord | 7 hours ago
It's been great, much more like HN used to be.
leephillips | 3 hours ago
The problem now is that so many articles with innocent titles are really about AI slop. The title might be “A Space Exploration Game” and the content is about some vibe-coded garbage.
ristomatti | an hour ago
gigatexal | 12 hours ago
Very interesting though. I don’t always read entire posts on blogs but this one I did. Lisp looks really interesting.
Zambyte | 10 hours ago
My long term vision is to make an Emacs implementation that is compatible only in philosophy. It would use Guile instead of Elisp, default to bindings that are more familiar to people coming from more modern systems, and would be built from the beginning with concurrency and graphics in mind. For now it remains a dream though.
grumpyprole | 10 hours ago
bergheim | 9 hours ago
krupan | 5 hours ago
I would guess you hadn't done as much emacs yak shaving as some of us other emacs users if the switch to VSCode was a simple one
cmrdporcupine | 9 hours ago
https://github.com/lem-project/lem
Zambyte | 4 hours ago
I want it to be as easy to make scripts to automate graphical applications as it is to automate textual ones in Emacs or shell scripts.
cmrdporcupine | 3 hours ago
As I said in my other comment in this topic I actually would love to see an arch where the UI portions are split up with a background daemon holding buffers, lisp execution etc and then IPC to frontend pieces for window management and buffer editing.
So window management can be done by ... a window manager, but with intelligent interaction to the editor pieces so you don't lose all the awesome emacs stuff.
EDIT: I would say however that something like lem is probably more amenable to that refactoring/restructuring than GNU emacs, which is a single-threaded monolith.
volemo | 8 hours ago
krupan | 6 hours ago
Zambyte | 4 hours ago
timonoko | 10 hours ago
timonoko | 9 hours ago
Try C-x 2, C-x 3 and C-x 4
the-lazy-guy | 6 hours ago
timonoko | 6 hours ago
krupan | 5 hours ago
It really speeds things up
nesarkvechnep | 9 hours ago
krupan | 5 hours ago
Pay08 | 4 hours ago
krupan | 5 hours ago
When we switched to using a tiling window manager for all our windows, we ran into a muscle memory problem. We were used to jumping between emacs windows/buffers using C-x o and C-x b and then without thinking about it we'd try and use the same keys to jump between i3/sway windows and of course it doesn't work. Or vice versa, trying to use i3/sway shortcuts to switch emacs windows/buffers.
To try and solve this problem I've been using less emacs windows and more i3/sway windows, so I can just use i3/sway keybindings everywhere, but emacs puts up some resistance to that. I like this approach
cmrdporcupine | 5 hours ago
I think the more elegant model would be for the window manager to own all the window management keybindings and then for there to be a background emacs-like process that owns the buffers (and the remaining keybindings) and can present them in standalone windows which the window manager presents. With some kind of IPC between the two for coordination, I guess.
I don't think GNU Emacs itself works well with this model tho. Its "server" mode is something else entirely.
9999gold | an hour ago
ristomatti | an hour ago
Recently I hacked together a Bash script for seamless navigation between i3 windows/tabs and Kitty windows/tabs using i3-msg, xdotool, jq, kitty's remote control sockets. Now if I only figured out how to add Helix windows into the mix...
Sometimes thinking outside the box helps. I've been able to unify a lot of keybindings by using a context aware remapping tool such as Keymapper (https://github.com/houmain/keymapper.
Pay08 | 3 hours ago
klik99 | 2 hours ago