Regarding the shell != terminal issue that is mentioned in several other comments... You've named a valid reason to blur the shell/terminal distinction. Nice.

Another would be: the ability to manipulate streams (or whatever you'd like to call the content that flows through pipes). As in, replay a stream into some new process without re-executing the original.

Another would be to fold terminal output.

Removing the shell/terminal distinction opens a door to a world of possibilities. But, the door closes behind you!

> Another would be: the ability to manipulate streams (or whatever you'd like to call the content that flows through pipes). As in, replay a stream into some new process without re-executing the original.

My terminal (https://github.com/sedwards2009/extraterm) can more or less do that now:

https://raw.githubusercontent.com/sedwards2009/extraterm/mas...

> Another would be to fold terminal output.

I did have something like that, but had to disable it. That feature will make a reappearance sometime as I want it for things like automatically collapsing command (e.g. compiler) output if terminals successfully. I only need to see the output of failed builds, for example.

So you'd use your mouse to cd into directories after you `ls`?

Moving you hand to the mouse, moving the mouse, clicking and clicking is quicker than `cd foo<TAB>`? Or just using a file manager and right click open terminal here?

I never said that. I just want the option. Normally I just `j` everywhere, but if I'm already mousing, let me click.