tab-rs VS endbasic

If you want to read some code, my project is tab-rs.

I myself use the many (alacritty) terminals + tiling WM solution at the moment (switching between i3wm and LeftWM) but it doesn't feel optimal. I always though tmux looked too involved to learn so I've been on the lookout for alternatives such as Wezterm (a terminal with built-in multiplexing), tab (a command line controlled multiplexer) and now zellij.

Plus a lot of cleanup in the tab-pty-process crate. It now exposes an interface similar to portable-pty, but with non-blocking file handles.

I ended up writing a terminal multiplexer because screen and tmux were too complicated to use. It has a built in fuzzy finder, stateless navigation, and YAML configs for persistent sessions: https://github.com/austinjones/tab-rs/

The way I typically unify messages is define an enum, and map/merge channel receivers. tokio-stream would probably work with these examples. Here's an example from a fuzzy-finder implementation: https://github.com/austinjones/tab-rs/blob/main/tab-command/src/service/terminal/fuzzy.rs#L332

Really nice cheat-sheet write up on screen & tmux & byobu.

Also worth checking out tabs-rs[1] which seems very well reviewed & recent.

Personally, I am a huge fan of dtach[2][3], which isn't a multiplexer, just a detachable proxier of terminal sessions. This let's me run a persistent vim session that I can reconnect to, and vim has however many terminals I need open in it. Vim does my multiplexing, dtach just allows me to make vim persistent. Very glad to have re-discovered dtach, to enable this workflow.

Notably dtach is very lightweight. Unlike tmux, it is not a virtual terminal. Upon reconnect to my vim session, I issue a control-l to refresh the screen. Dtach hasn't retained the screen state, isn't translating between terminfos. The one thing that can go wrong here is connecting from different terminals- few programs have a way to update the TERM setting once the program has launched.

[1] https://github.com/austinjones/tab-rs

[2] https://github.com/crigler/dtach

[3] https://wiki.archlinux.org/index.php/Dtach

The nice thing about that is that many other tools can work too. Someone using direnv can set properties that would be available in the hurl script. Likewise, someone using tab could have environment variables defined for their active tab that could be used. If you invent your own notion of an environment, you lose interop with a lot of other tools that target the standard environment.

Are you running v0.5.3? I just released a fix for a Kakoune issue that was caused by add-highlighter global/ number-lines -relative in kakrc. It sounds similar to what you described.

There are also some crazy ANSI sequences that cause the terminal emulator to write stdin - so applications can query the terminal state. Crazy stuff can happen when those sequences are copied from the scrollback buffer (which is why tab now filters them out).

I can confirm that writing a terminal is fun, for the reasons mentioned in the article: it’s easy to get “self-hosting”, but then the possibilities are endless :)

In my case, this was about creating the terminal for EndBASIC (https://www.endbasic.dev/). I wanted to mix text and graphics in the same console, so I had to ditch Xterm.js and create my own thing. It was really exciting to see graphics rendering mix with text “just fine” when I was able to render the first line.

I tried to set up a Raspberry Pi and configured it to boot into a simple window manager with DosBox full screen by default. I taught my kids to launch games within that and they learned the very basics… but it didn’t stick: they haven’t really gained any interest in how to do other stuff in the shell.

Anyway: check (my own) https://www.endbasic.dev/ which I’ve written precisely for the situation you describe :) You would actually have to /write/ the games first though!

Slightly buried: Apache 2.0, written in Rust, https://github.com/endbasic/endbasic/

Definitely an interesting attempt to cut through layers of abstraction and make something that lets people make the computer do useful/interesting things. No idea how well they realize that vision, of course, but good idea.

Or you could use https://www.endbasic.dev/

Somebody else brought it up in a separate comment, but because you specifically ask about the Raspberry, I'll mention EndBASIC (https://www.endbasic.dev/) here again :)

Supporting this platform has been a primary goal of mine, and in fact, one of the features (GPIO) only works on the Raspberry Pi today :) But there is a long road ahead. My vision is to create a minimal Linux image that boots straight into EndBASIC, and extend EndBASIC to give you more control of the Pi's hardware. The idea is to truly mimic the old C64 experience, but leveraging the power of modern hardware / infrastructure.

Agree with the author’s thesis of how the folks that “grew with computers” have an advantage over those approaching them now, in terms of understanding the inner workings. I’m not sure that this matters much in terms of solving actual problems though, which is probably a good thing.

But I somehow find it a little bit sad that this is the case, so I’ll plug my own https://www.endbasic.dev/ because it’s very fitting in this context :) I’ve been building it precisely as a way to understand everything that’s going on (although it’s still far from fulfilling that promise).

Also, buried in the article is a reference to the https://10print.org/ book. I recently came across it at HPB and it has been a pretty entertaining read. Couldn’t believe there was so much to write about such a simple little program!