How can I tell when non-interactive bash process is done executing a command?

Question

I'm trying to write a different shell command line interface where the input line stays at the top. Output of previous commands should stack below:

> INPUT LINE

LAST OUTPUT (of command 1)

PENULTIMATE OUTPUT (of command 2)

...

I want to use a non interactive bash process and would like to handle the insertion of the output into the process buffer via an output filter. Commands are send to the bash process via process-send-string.

As I tested the output of a command can arrive in chunks and I would like to know when all output has arrived because later I want to add features for which I need to know that (specifically I want to narrow the buffer after all output arrived).

So is it possible to determine when the the last chunk of output arrives the process filter?

Any answers explaining that this is totally the wrong way and what I should do instead are welcome, too!

Answer 1

One way I discovered is to construct the command send to the bash process in such a way that after the execution the bash process gets stopped:

(format "%s; kill -s STOP %s\n" command (process-id bash-process))

This way a process-sentinel can be used to detect when the execution is done and can enable the process again:

(defun sentinel (proc status)
  (when (string= status "stopped (signal)\n")
    ...
    (continue-process proc)))