IEX process finishes before executing the entire code-block and emacs seems to hang

[dhruv-dave]

Hello, I am facing an issue on my laptop where I have a code block I execute with the :session parameter and emacs seems to hang.
I have narrowed it down to what's causing the issue, but I cannot tell why it's causing the issue. Hoping someone can shed some light on this:

OS:

╰─ cat /etc/os-release 
PRETTY_NAME="Ubuntu 21.10"
NAME="Ubuntu"
VERSION_ID="21.10"
VERSION="21.10 (Impish Indri)"
VERSION_CODENAME=impish
ID=ubuntu
ID_LIKE=debian
HOME_URL="https://www.ubuntu.com/"
SUPPORT_URL="https://help.ubuntu.com/"
BUG_REPORT_URL="https://bugs.launchpad.net/ubuntu/"
PRIVACY_POLICY_URL="https://www.ubuntu.com/legal/terms-and-policies/privacy-policy"
UBUNTU_CODENAME=impish

EMACS Version:

"GNU Emacs 27.1 (build 1, x86_64-pc-linux-gnu, GTK+ Version 3.24.23, cairo version 1.16.0)
 of 2021-01-18, modified by Debian"

ORG Mode Version:

Org mode version 9.5.2 (9.5.2-gfbff08 @ /home/me/.emacs.d/elpa/org-9.5.2/)

Using edebug i have narrowed what works for me and what doesn't:

-- code-block

#+begin_src elixir :session testing
  1+1
#+end_src

-- what doesn't work:
I execute the source block (from inside the source block I run C-c C-c), and the buffer *testing* gets created and emacs hangs. I can get the control back by pressing C-g C-g.
Here is what I see in the *testing* buffer:

Interactive Elixir (1.13.0) - press Ctrl+C to exit (type h() ENTER for help)
iex(1)> IEx.configure(colors: [enabled: false])
:ok
iex(2)> 
Process testing finished

At this point, the *testing* buffer has no process running in it and I have to kill the buffer.

-- what works
I have narrowed down to the code causing the issue and "fixed" it. The fix is not ideal, but I am new to emacs and I am not sure why this "fixes" the issue. I changed the following lines in ob-elixir.el. With this change, the code block evaluates in the iex session and the results get printed back into the org file.

(defun org-babel-elixir-evaluate-session (session body)
  "Evaluate BODY in SESSION, i.e, the comint buffer."
  (let* ((buffer (format "*%s*" session))
         (send-wait (lambda ()
                      ;; (comint-send-input t nil) ;; this is causing the iex Process to exit early
                      (comint-send-input nil nil)  ;; setting the optional `no-newline` parameter to `nil` "fixes" the issue.
                      (sleep-for 0 5)))
         (insert-line (lambda (line)
                        (insert line "\n")
                        (funcall send-wait)))
         (lines `("IEx.configure(colors: [enabled: false])"
                  ,body
                  "IEx.configure(colors: [enabled: true])"))
         ;; set results (comint output)
         (results (org-babel-comint-with-output (buffer org-babel-elixir-hline t body)
                    (dolist (line lines)
                      (funcall insert-line line)))))
    ;; return results (convert to strings)
    (prin1-to-string (butlast (nthcdr 2 results) 3))))

Here is what I see in the org file:

#+begin_src elixir :session testing
  1+3
#+end_src

#+RESULTS:
| iex(10)> nil |
| iex(11)> 4   |
| iex(12)> nil |

At this point the *testing* buffer is still active, there is a iex process running in it and I can evaluate any previously evaluated code from the org file.

Also I found that in the function comint-send-input, if the no-newline parameter is set to t, then this is what causes the process to exit early:

;; line 1915
;; here, the `no-newline` is set to `t`
(let ((comint-input-sender-no-newline no-newline))
          (funcall comint-input-sender proc input))

Now the RESULTS section is not ideal, and I can see why the no-newline needs to be set to true.
Also, in the interactive elixir (iex) buffer, if i move the cursor to another position and evaluate the source block, the RESULTS section will print nil.

Please let me know if i can be help debug this issue further.

[lambdart]

Thanks for the report, I'll take a look ASAP.

[lambdart]

Hi @dhruv-dave, how are you?

I fix the problem (hope so), can you test it? :)

[dhruv-dave]

@lambdart thank you, I can test it out soon and get back to you!

[lambdart]

Maybe is necessary to add more filters to the replace-regex function.

[dhruv-dave]

@lambdart it solves the issue for me.
I do notice that when i execute the code for the very first time, it creates the session (in my case demo) buffer but doesn't populate the #RESULTS block/drawer. The consecutive execution of the source blocks also doesn't populate the #RESULTS drawer. If i follow the following steps, it works like a charm:

1. I open the buffer called demo
2. move the cursor to the end of the buffer
3. Execute the source block and the the #RESULTS block/drawer gets populated

[lambdart]

Thanks, I'll take a look on that 👍

[dhruv-dave]

thank you