use *-term instead of shell?

[brendan-r]

Firstly, thank you for this excellent package!

I am brand new to both emacs and elisp, so forgive me if this is a trivial PR (or obviously a bad idea), but would it be possible to extend the package to allow the evaluation of bash code in multi-term or ansi-term as opposed to shell?

Regardless, thanks for your work!

[kaz-yos]

That's an interesting idea. I'll look into that.

[kaz-yos]

I've created a version that parallels how the M-x shell version works, but it does not seem to work similarly. eval-in-repl relies on being able to switch to the REPL buffer and paste and execute the code in the script. It can paste the code into the *terminal* buffer, but execution does not work for some reason. The special behaviors that the *terminal* buffer has (keys behave differently) may be causing this.

You can try out the no-yet-functional partial implementation below if you are interested.

https://github.com/kaz-yos/eval-in-repl/blob/feature/term-support/eval-in-repl-term.el

[actondev]

That would be really useful, so I might take a look as well. @kaz-yos I see that the branch you linked is no longer available. In any case, do you happen to have some notes about noticable shell/term differences while you were working on the term support?

Looking forward to having shell & term support, would be really useful for devops coupled with https://github.com/diadochos/org-babel-eval-in-repl

From some search it seems like there's the following similar commands which should be selected depending on the shell or term mode

• comint-send-string and term-send-string
• comint-send-input and term-send-input

So I think having a var for preferring shell or term (to make term work with the existing eval-in-shell command) should be trivial. That said, I might take a look into it the following days.

@kaz-yos let me know what you think, and also if I you could write here in case you go on doing some work on this, to avoid doing duplicate work :)