Re: [Matlab-emacs-discuss] resend:matlab-shell-run-cell can only run one cell, narrow-to-region

[Uwe B. <ou...@ma...>]

>> On Sun, 12 Feb 2012 17:20:26 +0000, Eric Ludlam <Eri...@ma...> wrote:


Hi Eric,
   > Hi Uwe,

It seems an earlier message did not make it to the list.

   > I can't run matlab-shell at the moment to help answer the question,
   > but I'm confused as to what you are expecting.  In your example, of
   > the cursor is on line 2, run cell should only run x=8.  If your cursor
   > is on the line for y=91, it should only run that.  I tried the parsing
   > bit of matlab-shell-run-cell without executing the code, and it seemed
   > to work as expected for me.  Are you seeing some other behavior?


Yes but the reason is the behaviour of
matlab-shell-run-region. The issue is the cell line contains
naturally %% and then the current implementation of
matlab-shell-run-region just ignores the lines of that cell.

If I follow the advice as expressed in 
the message:  Bug in matlab-shell-run-region


,----
| Hello All,
| 
| I just bumped into something looking suspiciously like a bug.
| 
| symptoms:
| while running matlab-shell-run-region over matlab code including
| comments, all the code followed by the commented line is no executed.
| 
| causes:
| In the definition 'matlab-shell-run-region @ matlab.el, while creating
| the 'command binding, there is a 'while loop labeled as ;; HACK FOR
| NOSHOW whose action is to replace '\n' by ',' regardless of the present
| line being or not a comment.
| 
| aesthetic considerations:
| I do prefer to see the lines ran in the prompt as they were layered out
| by me in the code file. I can understand some people might wish to
| completely remove those lines from the matlab-shell buffer; that would
| certainly be another effort, different from simply concatenating
| everything in a big one-liner...
| 
| 
| the cure:
| Just comment the hack code...
| 
`----



then the problem which occurs in matlab-shell-run-cell
disappears!

That is the culprit is the 

the code 
		   ;; HACK FOR NOSHOW
 		   (while (string-match "\n" str)
 		     (setq str (replace-match ", " t t str)))

Which I commented out!

I strongly recommend to modify matlab-shell-run-region and
to remove that code or define a variable 

(defvar matlab-shell-run-no-show nil
*Variable which defines whether to replace '\n' by ','
regardless of the present line being or not a comment.)

So matlab-shell-run-region should look like:


(defun matlab-shell-run-region (beg end)
  "Run region from BEG to END and display result in MATLAB shell.
This command requires an active MATLAB shell."
  (interactive "r")
  (if (> beg end) (let (mid) (setq mid beg beg end end mid)))
  (let ((command (let ((str (concat (buffer-substring-no-properties beg end)
 				    "\n")))
 		   (while (string-match "\n\\s-*\n" str)
 		     (setq str (concat (substring str 0 (match-beginning 0))
 				       "\n"
 				       (substring str (match-end 0)))))
		   ;; HACK FOR NOSHOW
(if  matlab-shell-run-no-show 
 		   (while (string-match "\n" str)
 		     (setq str (replace-match ", " t t str))))
		   (setq str (concat str "\n"))
 		   str))
 	(msbn nil)
 	(lastcmd)
	(inhibit-field-text-motion t))
    (if (matlab-with-emacs-link)
	;; Run the region w/ Emacs Link
	(matlab-eei-eval-region beg end)

      (save-excursion
	(setq msbn (matlab-shell-buffer-barf-not-running))
	(set-buffer msbn)
	(if (not (matlab-on-prompt-p))
	    (error "MATLAB shell must be non-busy to do that"))
	;; Save the old command
	(beginning-of-line)
	(re-search-forward comint-prompt-regexp)
	(setq lastcmd (buffer-substring (point) (matlab-point-at-eol)))
	(delete-region (point) (matlab-point-at-eol))
	;; We are done error checking, run the command.
	(matlab-shell-send-string command)
	(insert lastcmd))
      (set-buffer msbn)
      (goto-char (point-max))
      (display-buffer msbn))
    ))


But I leave it up to you and others to decide.....

Uwe