Intriguing idea! I love the idea of having a database of the commands you’ve run and their output. I use tmux religiously, but it’s sometimes frustrating using a crappy interface to spelunk my terminal history and manually look for the various commands I’ve run amidst voluminous output. And of course it’ll lose that history when it runs out of buffer.
Question: You refer to this as a shell, but it seems more like a terminal emulator. Am I missing something?
A couple ideas for different ways to achieve the goal:
1. It seems like a web UI rather than a native app would be beneficial for various reasons (run anywhere, connect remotely, coded in the UI lingua franca, etc).
2. I could see value in separating the shell and the history browser. Like the shell is just your shell, and it can be used via whatever terminal emulator that people want to use, or even thru ssh. The only thing the shell does above and beyond any other shell is to pipe stuff to the database. If the user wants to look at history, then they go to the web UX (ala suggestion 1) to do that.
Doing this would make it more palatable for people to start using it because it wouldn’t change the feel of their daily experience, but in the background, they’d be getting the cool features.
There’s a lot of snap bashing in here, but one thing that seems really appealing is that snaps are run in sandboxes and protected by AppArmor profiles. I don’t know how good the implementation is, but sandboxing would seem to alleviate the risk that one of the hundreds of rando-authored packages that get installed on your system might decide to abuse root privileges during installation and quietly take over your system. Also, sandboxing your programs seems like a good idea in general, so they can’t all access each other’s data.
Question: You refer to this as a shell, but it seems more like a terminal emulator. Am I missing something?
A couple ideas for different ways to achieve the goal:
1. It seems like a web UI rather than a native app would be beneficial for various reasons (run anywhere, connect remotely, coded in the UI lingua franca, etc).
2. I could see value in separating the shell and the history browser. Like the shell is just your shell, and it can be used via whatever terminal emulator that people want to use, or even thru ssh. The only thing the shell does above and beyond any other shell is to pipe stuff to the database. If the user wants to look at history, then they go to the web UX (ala suggestion 1) to do that.
Doing this would make it more palatable for people to start using it because it wouldn’t change the feel of their daily experience, but in the background, they’d be getting the cool features.