Re: number of column of octave terminal

[Francesco Potorti`]

Thank you both for your explanations and interest.

I wrote:
   
   > How can  one make octave  change its idea of  terminal width?  I use
   > octave  as an emacs subprocess  on a character terminal.
   
   I made  the following experiments. If I  run octave in a shell buffer,
   then it seems that it takes all the columns. (I have 90).

Yes, it works for me also.  In fact, comint tells the process that it
runs on a dumb terminal with the correct number of columns.
   
   If I reshape the  emacs  window, then it   seems that  octave  doesn't
   change its idea of  the number of  columns  and it continues  to print
   matrices as if the number of columns were 90.

That is perhaps because the SIGWINCH does not reach the octave
process?  Or even if it reaches it, readline does not handle it?
   
   On the other hand, if I use inferior-octave (from Kurt Hornik and jwe)
   then octave takes the full  width of the window  

I use the following home-made function which I wrote long ago and is
now modified to use comint, and now I get the right window size:

(defun octave (rawprefix)
  (interactive "P")
  (if (and (get-buffer "octave") (null rawprefix))
      (switch-to-buffer "octave")
    (let ((comint-scroll-to-bottom-on-input 'this)
          (comint-scroll-to-bottom-on-output nil)
          (comint-scroll-show-maximum-output t)
          (buffer (generate-new-buffer "octave")))
      (comint-exec buffer "octave" "octave" nil nil)
      (switch-to-buffer buffer)
      (shell-mode)
      (cd "~/math/workarea"))))

   Do you use inferior-octave? 

I do not, because on this system (Debian Linux, emacs 19.34), M-x
run-octave hangs (or loops).  I have not had the time to investigate
the problem, so I stick with my home-made function for the moment.

Thanks again.
F.