@mwl So I've now read the book. Actually I've worked through all the examples except the most complicated regexes (which was my threshold for "too lazy to type in"). Even in the simple examples that I already knew how they worked, I had fun.

I knew how ed(1) works, but I have now a much more systematic knowledge. Worth the few bucks, definitely worth the time.

And also ... I got this feeling for how people used to work with ed, how working with it actually works. I know I could do this.

@0x4d6165 there is a case to be made for writing a first draft in ed (seriously!) I accidentally figured this out while using ed in part of a project of mine as an experiment.

Ed's insert mode is actually pretty ideal for getting into a flow state. You commit things one line at a time. The line can be as long as you want. (I think) you can use emacs/vi key bindings with rlwrap to edit the line. But once you hit enter, the line is in, and you're onto the next thought.

@jack @kkarhan @mwl Given the current zeitgeist someone, somewhere, RIGHT NOW, is working to embed Claude or Gemini in the ed(1) command prompt. And they're going to end up giving it a dependencies on Gtk and react.js.

https://mastodon.sdf.org/@jack/116190242608001379

For whomever needs to hear this: you are not alone. I am not vibecoding any of the software I write. I write it all by hand. I have leveled up to using ed(1) on a glass TTY rather than a hardcopy-output device. I am modernizing my stacks and using languages with excellent compilers like C89. I think about how to do more with less. I am trying to combine the best human-written libraries and modules and assemble them with minimal boilerplate. I enjoy reading manuals and references. I believe in robust, secure, human-written software.

@mntmn