Re: the dumb Emacs terminal

[John Mastro]

Emanuel Berg <embe8573@student.uu.se> wrote:
>> M-x shell isn't really a terminal (emulator)
>> at all, which is also why you can't run e.g.
>> htop there.
>
> Right, you can't use even more basic stuff like
> 'clear'. Is it just able to output new lines,
> not change the state of what has already
> been outputted?

Yes, pretty much - it just sends a line of input and then prints the
output.

>> I nonetheless prefer M-x shell over Emacs's
>> terminal emulators.
>
> Why go into the small workshop, when there is
> a big?

I like the simplicity and that I can work in/with the *shell* buffer
just like any "normal" buffer. I do, of course, sometimes use programs
that don't work there, but it's rare enough that I don't mind switching
to a separate terminal emulator on those occasions.

Many of the things I previously would have done in a terminal emulator
that wouldn't have worked well in M-x shell I now do in various Emacs
modes anyway. (Mail, some version control tasks, reading man pages,
etc.)

> Only a lot of shortcuts including M-x seem to
> be shadowed by the shell, that must be dealt
> with as I seldom or never use them with the
> shell but with Emacs I do it all the time
> obviously. So it should be mergable that way,
> God willing.

Both term and ansi-term make a distinction between "line mode" and "char
mode", which send input to the shell a line at a time or a character at
a time respectively. They start out in char mode by default; you can
switch to line mode with C-c C-j and back to char mode with C-c C-k.
Line mode is more similar to M-x shell and using it, when possible,
helps close some of the distance.

        John