matlab-emacs-discuss — Discuss Emacs/MATLAB integration

Re: [Matlab-emacs-discuss] Building matlab-mode on Windows

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

>>> "VN" == Vladimir Nikishkin <loc...@gm...> writes:

   > Hello, everyone
   > I wanted to install matlab-mode on windows, but the INSTALL file is a
   > bit confusing.

   > Firstly, I don't have a standard `make' on Windows, and I don't know
   > which one would work best. (I guess, the one with which Emacs itself
   > is built on Windows, but I don't know exactly how this is done.)

   > Secondly, I tried to just simply copy the source to
   > .emacs.d/elpa/matlab-mode/,

Why do you do this????

Why don't you use the package manager and install it in clean way?

Re: [Matlab-emacs-discuss] Building matlab-mode on Windows

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

>>> "VN" == Vladimir Nikishkin <loc...@gm...> writes:

   > Simple answer: I don't have an internet access on this machine.

But you do have access to the source code in sourceforge? That is really
odd. Unfortunately I am on Linux, so I don't have experience with make
on windows. You might try to install cygwin, if you have access to that
site, of course.

Re: [Matlab-emacs-discuss] Building matlab-mode on Windows

[Vladimir N. <loc...@gm...>]

Simple answer: I don't have an internet access on this machine.

Uwe Brauer <ou...@ma...> 于 2019年2月20日周三 17:13写道：

> >>> "VN" == Vladimir Nikishkin <loc...@gm...> writes:
>
>    > Hello, everyone
>    > I wanted to install matlab-mode on windows, but the INSTALL file is a
>    > bit confusing.
>
>    > Firstly, I don't have a standard `make' on Windows, and I don't know
>    > which one would work best. (I guess, the one with which Emacs itself
>    > is built on Windows, but I don't know exactly how this is done.)
>
>    > Secondly, I tried to just simply copy the source to
>    > .emacs.d/elpa/matlab-mode/,
>
> Why do you do this????
>
> Why don't you use the package manager and install it in clean way?
>

[Matlab-emacs-discuss] Building matlab-mode on Windows

[Vladimir N. <loc...@gm...>]

Hello, everyone

I wanted to install matlab-mode on windows, but the INSTALL file is a
bit confusing.

Firstly, I don't have a standard `make' on Windows, and I don't know
which one would work best. (I guess, the one with which Emacs itself
is built on Windows, but I don't know exactly how this is done.)

Secondly, I tried to just simply copy the source to
.emacs.d/elpa/matlab-mode/,
add it to the 'package-selected-packages variable in .emacs
and issue
(require 'matlab-load), but it still doesn't work

Perhaps, matlab-load is only created during the build process?
Can someone help me with this?

I'm using Emacs 26.1

-- 
Yours sincerely, Vladimir Nikishkin

[Matlab-emacs-discuss] Adding flycheck mlint support

[W. R. L. <wry...@gm...>]

Hi all,

I've been playing with using flycheck to highlight MLINT warnings/errors
rather than the existing code in mlint.el. flycheck is a popular and robust
package for code checkers and it would be great to support in matlab-mode.
So I'm writing to propose adding it to the matlab.el distribution.

The code I've been using is at the end of this message. Some open questions
I have:

1. What's the best way to integrate with matlab-mode? Right now I just turn
off the current mlint highlighting in the setup function:

(defun flycheck-matlab-mlint-setup ()
  "Set up Flycheck MATLAB mlint.


Adds `matlab-mlint' to `flycheck-checkers'.


NOTE: This disables showing warnings with the matlab-mode mlint


integration"
  (interactive)
  (add-to-list 'flycheck-checkers 'matlab-mlint)
  (eval-after-load 'matlab
    '(setq matlab-show-mlint-warnings nil)))

2. How could we integrate this with the mlint.el heuristics for finding the
mlint executable? Right now a user has to
set flycheck-matlab-mlint-executable. Maybe this could be more automated.

Aside from that, I've really liked flycheck integration as it provides many
niceties and will continue to provide more as the underlying infrastructure
evolves.

Thanks,
Ryan

Code:

;;; flycheck-matlab-mlint.el --- Flycheck: MATLAB support  -*-
lexical-binding: t; -*-

;;; Commentary:

;; Provides support for MATLAB code checking with mlint and flycheck
;; that integrates with matlab-mode
;;
;; Usage:
;;
;;     (eval-after-load 'flycheck
;;       '(add-hook 'flycheck-mode-hook #'flycheck-matlab-mlint-setup))

;;; Code:

(require 'flycheck)

(flycheck-define-checker matlab-mlint
  "A MATLAB checker using MATLAB mlint"
  :command ("mlint" "-id" "-all" source-original)
  ;; Example mlint messages.
  ;; L 588 (C 46-49): LOAD: To avoid conflicts with functions on the path,
specify variables to load from file.
  ((warning line-start "L " line " (C " column "-" column "): " (id (*
alnum)) ":" (message))
   (warning line-start "L " line " (C " column "): " (id (* alnum)) ":"
(message))
   )
  :modes (matlab-mode)
  :predicate (lambda () (flycheck-buffer-saved-p)))

;;;###autoload
(defun flycheck-matlab-mlint-setup ()
  "Set up Flycheck MATLAB mlint.

Adds `matlab-mlint' to `flycheck-checkers'.

NOTE: This disables showing warnings with the matlab-mode mlint
integration"
  (interactive)
  (add-to-list 'flycheck-checkers 'matlab-mlint)
  (eval-after-load 'matlab
    '(setq matlab-show-mlint-warnings nil)))

(provide 'flycheck-matlab-mlint)

;;; flycheck-matlab-mlint.el ends here

Re: [Matlab-emacs-discuss] error when running system command

[Ondra K. <ond...@gm...>]

Dear all,

the reported problem does not seem to concern matlab-shell at all. So my
email can be ignored here.

I found out the similar problem happens in 2018b but for legend command
when run in matlab-shell, and then I found out that this problem is also in
2018a if run from terminal. It does not depend on matlab-shell in emacs. It
seems to depend on the order i ran different versions of Matlab.

It rather seems that the problem might be connected with three Matlab
installations, so i will investigate. Maybe my ~/pathdef.m is not loading
all the paths because it gets overwritten by another matlab? Anyway, i
apologize for disturbing a wrong list.

0.


On Tue, Jan 29, 2019 at 7:00 PM Ondra Kamenik <ond...@gm...>
wrote:

> Dear all,
>
> matlab-mode.el is great! I have been using it for years with great
> satisfaction. Thank you for your work.
>
> I installed today 2018a and also 2018b. My previous version was 2015b. If
> I run matlab in Emacs matlab-shell-mode than I get the following error when
> calling "system('ls')" for example:
>
> >> format compact
> >> system('ls')
> Undefined variable "matlab" or class
> "matlab.io.internal.utility.convertStringsToChars".
> Error in dos (line 65)
> [varargin{1:nargin}] =
> matlab.io.internal.utility.convertStringsToChars(varargin{:});
> >> usejava('jvm')
> ans =
>   logical
>    1
> >> ver('MATLAB')
>
> -----------------------------------------------------------------------------------------------------
> MATLAB Version: 9.4.0.813654 (R2018a)
> MATLAB License Number: 40706422
> Operating System: Linux 4.19.13-300.fc29.x86_64 #1 SMP Sat Dec 29 22:54:28
> UTC 2018 x86_64
> Java Version: Java 1.8.0_144-b01 with Oracle Corporation Java HotSpot(TM)
> 64-Bit Server VM mixed mode
>
> -----------------------------------------------------------------------------------------------------
> MATLAB                                                Version 9.4
>  (R2018a)
>
> If I run the same in matlab-shell for R2018b, i don't have any problem. If
> i run matlab 2018a from terminal with "-nodesktop", i don't have any
> problem either. The only problem is a combination of 2018a and
> matlab-shell. I checked outputs of matlab -n and matlab -e run from
> terminal and from matlab-shell, but i did not find any difference which
> could cause it.
>
> Many thanks for any leaders. Unfortunately, i cannot replace 2018a with
> 2018b as we have a lot of codes which still need be ported to 2018b.
>
> 0.
>

[Matlab-emacs-discuss] error when running system command

[Ondra K. <ond...@gm...>]

Dear all,

matlab-mode.el is great! I have been using it for years with great
satisfaction. Thank you for your work.

I installed today 2018a and also 2018b. My previous version was 2015b. If I
run matlab in Emacs matlab-shell-mode than I get the following error when
calling "system('ls')" for example:

>> format compact
>> system('ls')
Undefined variable "matlab" or class
"matlab.io.internal.utility.convertStringsToChars".
Error in dos (line 65)
[varargin{1:nargin}] =
matlab.io.internal.utility.convertStringsToChars(varargin{:});
>> usejava('jvm')
ans =
  logical
   1
>> ver('MATLAB')
-----------------------------------------------------------------------------------------------------
MATLAB Version: 9.4.0.813654 (R2018a)
MATLAB License Number: 40706422
Operating System: Linux 4.19.13-300.fc29.x86_64 #1 SMP Sat Dec 29 22:54:28
UTC 2018 x86_64
Java Version: Java 1.8.0_144-b01 with Oracle Corporation Java HotSpot(TM)
64-Bit Server VM mixed mode
-----------------------------------------------------------------------------------------------------
MATLAB                                                Version 9.4
 (R2018a)

If I run the same in matlab-shell for R2018b, i don't have any problem. If
i run matlab 2018a from terminal with "-nodesktop", i don't have any
problem either. The only problem is a combination of 2018a and
matlab-shell. I checked outputs of matlab -n and matlab -e run from
terminal and from matlab-shell, but i did not find any difference which
could cause it.

Many thanks for any leaders. Unfortunately, i cannot replace 2018a with
2018b as we have a lot of codes which still need be ported to 2018b.

0.

Re: [Matlab-emacs-discuss] Font-lock breakage?

[Eric L. <Eri...@ma...>]

I tried the case with:

% foobar 1

and it font-locked ok for me.

The comment/string highlighter won't highlight areas that are already highlighted by something else.  Do you have any other font-lock trickery going on that might be highlighting numbers before the matlab lockers start up?

Eric

From: ole petter orhagen <ol...@ho...>
Sent: Wednesday, October 24, 2018 12:18 PM
To: mat...@li...
Subject: Re: [Matlab-emacs-discuss] Font-lock breakage?


lol, looking a little closer at emacs regexps it seems this should match numerals.



Then I'm afraid I'm out of ideas on how to fix this in a nippy

________________________________
Fra: ole petter orhagen <ol...@ho...<mailto:ol...@ho...>>
Sendt: onsdag 24. oktober 2018 18.14.41
Til: mat...@li...<mailto:mat...@li...>
Emne: Re: [Matlab-emacs-discuss] Font-lock breakage?




________________________________
Fra: ole petter orhagen <ol...@ho...<mailto:ol...@ho...>>
Sendt: onsdag 24. oktober 2018 18.04
Til: mat...@li...<mailto:mat...@li...>
Emne: Font-lock breakage?


Hi,



It seems to me that font-locking is somewhat broken. I might be wrong however. The issue seems to be:



% foobar



Works just fine



% foobar1



Also works



However



% foobar 1



or any other number for that matter - will break font-locking.



However, looking at the function 'matlab-font-lock-comment-match', and the regex '\\(%[^%\n]*\\)', this will not match numerals.



Is this a conscious design decition?



Cheers

-O

Re: [Matlab-emacs-discuss] Font-lock breakage?

[ole p. o. <ol...@ho...>]

lol, looking a little closer at emacs regexps it seems this should match numerals.


Then I'm afraid I'm out of ideas on how to fix this in a nippy

________________________________
Fra: ole petter orhagen <ol...@ho...>
Sendt: onsdag 24. oktober 2018 18.14.41
Til: mat...@li...
Emne: Re: [Matlab-emacs-discuss] Font-lock breakage?




________________________________
Fra: ole petter orhagen <ol...@ho...>
Sendt: onsdag 24. oktober 2018 18.04
Til: mat...@li...
Emne: Font-lock breakage?


Hi,


It seems to me that font-locking is somewhat broken. I might be wrong however. The issue seems to be:


% foobar


Works just fine


% foobar1


Also works


However


% foobar 1


or any other number for that matter - will break font-locking.


However, looking at the function 'matlab-font-lock-comment-match', and the regex '\\(%[^%\n]*\\)', this will not match numerals.


Is this a conscious design decition?


Cheers

-O

Re: [Matlab-emacs-discuss] Font-lock breakage?

[ole p. o. <ol...@ho...>]

________________________________
Fra: ole petter orhagen <ol...@ho...>
Sendt: onsdag 24. oktober 2018 18.04
Til: mat...@li...
Emne: Font-lock breakage?


Hi,


It seems to me that font-locking is somewhat broken. I might be wrong however. The issue seems to be:


% foobar


Works just fine


% foobar1


Also works


However


% foobar 1


or any other number for that matter - will break font-locking.


However, looking at the function 'matlab-font-lock-comment-match', and the regex '\\(%[^%\n]*\\)', this will not match numerals.


Is this a conscious design decition?


Cheers

-O

Re: [Matlab-emacs-discuss] Patch for matlab.el to support string

[Eric L. <Eri...@ma...>]

Thanks Uwe!

Eric

-----Original Message-----
From: Uwe Brauer <ou...@ma...> 
Sent: Saturday, September 29, 2018 9:53 AM
To: Eric Ludlam <Eri...@ma...>
Cc: mat...@li...
Subject: Re: [Matlab-emacs-discuss] Patch for matlab.el to support string

>>> "Eric" == Eric Ludlam <Eri...@ma...> writes:

   > The attached patch adds support for Strings in matlab-mode.

Applied to the sf and github repo. Thanks Eric

Re: [Matlab-emacs-discuss] Patch for matlab.el to support string

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

>>> "Eric" == Eric Ludlam <Eri...@ma...> writes:

   > The attached patch adds support for Strings in matlab-mode.

Applied to the sf and github repo. Thanks Eric

[Matlab-emacs-discuss] Patch for matlab.el to support string

[Eric L. <Eri...@ma...>]

The attached patch adds support for Strings in matlab-mode.

Previously, character vectors such as:

a = 'char vector'
a = 'char '' vector'

were supported.   This patch enables

b = "string scalar"
b = "string "" scalar"

to also be highlighted.

There is one bug remaining which is that:

a = 'unterminated '' string

used to work, but now fails for char vector and string.    Unterminated strings still work for:

a = 'unterminated string

cases where there is no '' or "" in the string.

Enjoy
Eric

Re: [Matlab-emacs-discuss] executing matlab code in org file without copying code to the output buffer

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

>>> "LS" == LS CRT Sekretariat <k.s...@b-...> writes:

   > Hi
   > executing the matlab code (see below)

   > #+begin_src matlab :session :results output
   > clear
   > a=1;
   > i=1
   > #+end_src

   > #+RESULTS:
   > : clear
   > : a=1;
   > : i=1
   > :
   > : i =
   > :
   > :      1

   > the code is copied into the results buffer, which is not what I want.
   > Using octave instead of matlab will yield the correct output from stdout.
   > Any hint or help is greatly appreciated.

Hi 

The syntax you are using, seems to me a question to org mode, not to
matlab mode.


 Emacs matlab mode provides a mayor mode for .m files for editing it
 also provides a command shell written in lisp which allows you to
 display the results in a separate buffer, precisely what you are
 looking for. But it is different mayor mode.

I asked some time ago the inverse question, I wanted to use the lisp
matlab shell in org mode, since I wanted to *insert* the result of the
matlab calculations in the same mode I have the matlab commands. You
want the opposite.

So either, 

    1. Use the matlab emacs mode.

    2. Ask in the org mode list.


Or what do I miss?

Uwe Brauer 

[Matlab-emacs-discuss] executing matlab code in org file without copying code to the output buffer

[LS C. S. <k.s...@b-...>]

Hi

executing the matlab code (see below)

#+begin_src matlab :session :results output
clear
a=1;
i=1
#+end_src

#+RESULTS:
: clear
: a=1;
: i=1
:
: i =
:
:      1

the code is copied into the results buffer, which is not what I want.
Using octave instead of matlab will yield the correct output from stdout.
Any hint or help is greatly appreciated.

Kind regards

Klaus

Re: [Matlab-emacs-discuss] org mode: execute matlab code and insert it, the matlab shell, matlab live editor, jupiter notebook

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

>>> "Eric" == Eric Ludlam <Eri...@ma...> writes:

Hi Eric
   > Hi Uwe,

Thanks for your answer.
   > I did see this on the list the first time, and I spent a little
   > time on it. I was surprised to see the result from using
   > matlab-shell-collect-command-output cropping as you suggested.
   > There are a few filters for things like errors, anchors, and other
   > misc text to hide that I played with, but I found nothing that
   > matched \end at that time so I didn't end up replying as I hadn't
   > found anything. There may be something else with the special MATLAB
   > end keyword I missed. Process filters are a challenge to debug,
   > particularly when there is both the matlab filter plus the gud
   > filter interacting which I'm just not that familiar with anymore.

   > If you start in gud-matlab-marker-filter, you can see some of the
   > stuff it filters, such as escape codes, error, opentoline, etc.
   > Anyway, I didn't see anything there.

   > matlab-shell-collect-command-output perhaps could cheat, and
   > disable the process filter near the loop that accepts process
   > output. For example:

   > (condition-case nil
   >      (set-process-filter (get-buffer-process (current-buffer))
   > 			      nil)
   >      ;; loop is here
   >   (error nil))
   > (set-process-filter (get-buffer-process (current-buffer))
   > 			      'gud-filter)

Do you suggest, that I should give this modification a try?


Meanwhile I got  the python jupyther client, matlab kernel 
and emacs ob-ipython to work.


This is much faster than John's solution but slower than the emacs lisp
shell!!


   > but that's a pretty big hammer. I guess it would tell us if the
   > filter is the problem. If it is in this loop, there may be an issue
   > in prompt detection. The bit that cuts out the string to return is
   > pretty simple. Perhaps it is being confused by newlines or
   > something.

Maybe in my vacation I will play around a bit. There seems not much
interest in the matlab list nor in the org mailing list about this issue
though.

Regards

Uwe 

Re: [Matlab-emacs-discuss] org mode: execute matlab code and insert it, the matlab shell, matlab live editor, jupiter notebook

[Eric L. <Eri...@ma...>]

Hi Uwe,

I did see this on the list the first time, and I spent a little time on it.  I was surprised to see the result from using matlab-shell-collect-command-output cropping as you suggested.  There are a few filters for things like errors, anchors, and other misc text to hide that I played with, but I found nothing that matched \end at that time so I didn't end up replying as I hadn't found anything.  There may be something else with the special MATLAB end keyword I missed.  Process filters are a challenge to debug, particularly when there is both the matlab filter plus the gud filter interacting which I'm just not that familiar with anymore.

If you start in gud-matlab-marker-filter, you can see some of the stuff it filters, such as escape codes, error, opentoline, etc.   Anyway, I didn't see anything there.

matlab-shell-collect-command-output perhaps could cheat, and disable the process filter near the loop that accepts process output.   For example:

(condition-case nil
     (set-process-filter (get-buffer-process (current-buffer))
			      nil)
     ;; loop is here
  (error nil))
(set-process-filter (get-buffer-process (current-buffer))
			      'gud-filter)

but that's a pretty big hammer.  I guess it would tell us if the filter is the problem.   If it is in this loop, there may be an issue in prompt detection.  The bit that cuts out the string to return is pretty simple.  Perhaps it is being confused by newlines or something.

Eric


-----Original Message-----
From: Uwe Brauer [mailto:ou...@ma...] 
Sent: Monday, July 02, 2018 4:45 AM
To: matlab-emacs-discuss <mat...@li...>
Cc: John Ciolfi <Joh...@ma...>; Eric Ludlam <eri...@gm...>
Subject: [Matlab-emacs-discuss] org mode: execute matlab code and insert it, the matlab shell, matlab live editor, jupiter notebook


Hi

I am not sure whether this mail made to the list, I never saw it. I also cc this message to John and Eric since they might know more about the matlab shell.

One of new features in matlab is the matlab live editor, which allows to mix documentation, matlab code and, most importantly, the results of executed matlab code. 

It seems that the jupiter notebooks should have a similar feature, but I have not tried it out.

The question is whether there is a similar solution using emacs. 

And the answer to this is yes, but it is, at the moment, quite slow, at least with recent matlab releases.

The point is to use org mode.

Org mode is a mayor mode which becomes increasing popular since it ships with an amazing amount of features.

One of them is to execute code and insert the result in the org buffer, formatted if wished.
Now matlab is not really supported natively, but John Kitchin posted a while ago a nice solution on the org mailing list.

(defun org-babel-execute:matlab-john-org (body params)
  (interactive "P")
  (let* ((current-file (buffer-file-name)) 
	 (code (org-element-property :value (org-element-context)))
	 (result-params (cdr (assoc :result-params params)))
	 m-file
	 md5-hash)
    (with-temp-buffer 
      (insert code)
      (setq md5-hash (md5 (buffer-string))
	    mbuffer (format "*m-%s*" md5-hash)
	    m-file (format "m-%s.m" md5-hash)))
    ;; create the file to run
    (with-temp-file m-file 
      (insert code))
    (let ((results
           (shell-command-to-string
		    (concat
		     "/usr/local/bin/matlab "
		     "-nodesktop <"
;		     "-nodesktop -nojvm -nosplash -nodisplay <"
;		     "-nodesktop -nojvm -nosplash  <"
		     m-file))))
      (delete-file m-file)
      (when results
	;; strip out >>
	(setq results (replace-regexp-in-string ">> " "" results))
	;; remove first 10 lines that are the header.
        ;; matlab license seem to expire soon, so 5 warning lines are added
        ;; change first 10 to first 15 lines
	(setq results (mapconcat 'identity
				 (nthcdr 10 (split-string results "\n"))
				 "\n")))
      (org-babel-result-cond result-params
	results))))

There is a problem with his code, however: It starts every time it is called the whole matlab engine, making it quite slow.

So I thought of running the emacs matlab shell running and use it instead. So I came up with the following

(defun org-babel-execute:matlab-john-shell (body params)
  (interactive "P")
  (let* ((current-file (buffer-file-name)) 
         (code (org-element-property :value (org-element-context)))
         (result-params (cdr (assoc :result-params params))))
    (let ((results
           (matlab-shell-collect-command-output code)))
      (org-babel-result-cond result-params
        results))))



That works, however take the following example
#+begin_src matlab :results output latex :exports results :wrap latex
X=[1,2,3,4,5,6,7];
p=[1/7 1/7 1/7 1/7 1/7 1/7 1/7];
E=X*p';
x2=X.*X;
E2=x2*p';
V=E2-E^2;
disp('\begin{align}')
fprintf('E[X]&=%g\\\\ \n', E)
fprintf('E^2[X]&=%g\\\\\n', E2)
fprintf('V[X]&=%g\n', V)
disp('\end{align}')
disp('\end{align}')
#+end_src

By code only extracts:
#+RESULTS:
#+BEGIN_latex
>> >> p=[1/7 1/7 1/7 1/7 1/7 1/7 1/7];
>> >> E=X*p';
>> >> x2=X.*X;
\begin{align}
E[X]&=4\\ 
E^2[X]&=20\\
V[X]&=4
#+END_latex


But not the last 2 lines of the original matlab code.

Anybody with a better  know of the matlab-shell can explain that to me
please? In order to test the code I should send a bunch of more
functions, which I attach for convince.


Regards

Uwe Brauer

[Matlab-emacs-discuss] org mode: execute matlab code and insert it, the matlab shell, matlab live editor, jupiter notebook

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

Hi

I am not sure whether this mail made to the list, I never saw it. I also
cc this message to John and Eric since they might know more about the
matlab shell.

One of new features in matlab is the matlab live editor, which allows to
mix documentation, matlab code and, most importantly, the results of
executed matlab code. 

It seems that the jupiter notebooks should have a similar feature, but I
have not tried it out.

The question is whether there is a similar solution using emacs. 

And the answer to this is yes, but it is, at the moment, quite slow, at
least with recent matlab releases.

The point is to use org mode.

Org mode is a mayor mode which becomes increasing popular since it ships
with an amazing amount of features.

One of them is to execute code and insert the result in the org buffer,
formatted if wished.
Now matlab is not really supported natively, but John Kitchin posted a
while ago a nice solution on the org mailing list.

(defun org-babel-execute:matlab-john-org (body params) 
  (interactive "P")
  (let* ((current-file (buffer-file-name)) 
	 (code (org-element-property :value (org-element-context)))
	 (result-params (cdr (assoc :result-params params)))
	 m-file
	 md5-hash)
    (with-temp-buffer 
      (insert code)
      (setq md5-hash (md5 (buffer-string))
	    mbuffer (format "*m-%s*" md5-hash)
	    m-file (format "m-%s.m" md5-hash)))
    ;; create the file to run
    (with-temp-file m-file 
      (insert code))
    (let ((results
           (shell-command-to-string
		    (concat
		     "/usr/local/bin/matlab "
		     "-nodesktop <"
;		     "-nodesktop -nojvm -nosplash -nodisplay <"
;		     "-nodesktop -nojvm -nosplash  <"
		     m-file))))
      (delete-file m-file)
      (when results
	;; strip out >>
	(setq results (replace-regexp-in-string ">> " "" results))
	;; remove first 10 lines that are the header.
        ;; matlab license seem to expire soon, so 5 warning lines are added
        ;; change first 10 to first 15 lines
	(setq results (mapconcat 'identity
				 (nthcdr 10 (split-string results "\n"))
				 "\n")))
      (org-babel-result-cond result-params
	results))))

There is a problem with his code, however: It starts every time it is
called the whole matlab engine, making it quite slow.

So I thought of running the emacs matlab shell running and use it
instead. So I came up with the following

(defun org-babel-execute:matlab-john-shell (body params) 
  (interactive "P")
  (let* ((current-file (buffer-file-name)) 
         (code (org-element-property :value (org-element-context)))
         (result-params (cdr (assoc :result-params params))))
    (let ((results
           (matlab-shell-collect-command-output code)))
      (org-babel-result-cond result-params
        results))))



That works, however take the following example
#+begin_src matlab :results output latex :exports results :wrap latex
X=[1,2,3,4,5,6,7];
p=[1/7 1/7 1/7 1/7 1/7 1/7 1/7];
E=X*p';
x2=X.*X;
E2=x2*p';
V=E2-E^2;
disp('\begin{align}')
fprintf('E[X]&=%g\\\\ \n', E)
fprintf('E^2[X]&=%g\\\\\n', E2)
fprintf('V[X]&=%g\n', V)
disp('\end{align}')
disp('\end{align}')
#+end_src

By code only extracts:
#+RESULTS:
#+BEGIN_latex
>> >> p=[1/7 1/7 1/7 1/7 1/7 1/7 1/7];
>> >> E=X*p';
>> >> x2=X.*X;
\begin{align}
E[X]&=4\\ 
E^2[X]&=20\\
V[X]&=4
#+END_latex


But not the last 2 lines of the original matlab code.

Anybody with a better  know of the matlab-shell can explain that to me
please? In order to test the code I should send a bunch of more
functions, which I attach for convince.


Regards

Uwe Brauer

[Matlab-emacs-discuss] org mode: execute matlab code and insert it, the matlab shell

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

Hi

Org mode is a mayor mode which becomes increasing popular since it ships
with an amazing amount of features.

One of them is to execute code and insert the result in the org buffer,
formatted if wished.
Now matlab is not really supported natively, but John Kitchin posted a
while ago a nice solution on the org mailing list.

(defun org-babel-execute:matlab-john-org (body params) 
  (interactive "P")
  (let* ((current-file (buffer-file-name)) 
	 (code (org-element-property :value (org-element-context)))
	 (result-params (cdr (assoc :result-params params)))
	 m-file
	 md5-hash)
    (with-temp-buffer 
      (insert code)
      (setq md5-hash (md5 (buffer-string))
	    mbuffer (format "*m-%s*" md5-hash)
	    m-file (format "m-%s.m" md5-hash)))
    ;; create the file to run
    (with-temp-file m-file 
      (insert code))
    (let ((results
           (shell-command-to-string
		    (concat
		     "/usr/local/bin/matlab "
		     "-nodesktop <"
;		     "-nodesktop -nojvm -nosplash -nodisplay <"
;		     "-nodesktop -nojvm -nosplash  <"
		     m-file))))
      (delete-file m-file)
      (when results
	;; strip out >>
	(setq results (replace-regexp-in-string ">> " "" results))
	;; remove first 10 lines that are the header.
        ;; matlab license seem to expire soon, so 5 warning lines are added
        ;; change first 10 to first 15 lines
	(setq results (mapconcat 'identity
				 (nthcdr 10 (split-string results "\n"))
				 "\n")))
      (org-babel-result-cond result-params
	results))))

There is a problem with his code, however: It starts every time it is
called the whole matlab engine, making it quite slow.

So I thought of running the emacs matlab shell running and use it
instead. So I came up with the following

(defun org-babel-execute:matlab-john-shell (body params) 
  (interactive "P")
  (let* ((current-file (buffer-file-name)) 
         (code (org-element-property :value (org-element-context)))
         (result-params (cdr (assoc :result-params params))))
    (let ((results
           (matlab-shell-collect-command-output code)))
      (org-babel-result-cond result-params
        results))))



That works, however take the following example
#+begin_src matlab :results output latex :exports results :wrap latex
X=[1,2,3,4,5,6,7];
p=[1/7 1/7 1/7 1/7 1/7 1/7 1/7];
E=X*p';
x2=X.*X;
E2=x2*p';
V=E2-E^2;
disp('\begin{align}')
fprintf('E[X]&=%g\\\\ \n', E)
fprintf('E^2[X]&=%g\\\\\n', E2)
fprintf('V[X]&=%g\n', V)
disp('\end{align}')
disp('\end{align}')
#+end_src

By code only extracts:
#+RESULTS:
#+BEGIN_latex
>> p=[1/7 1/7 1/7 1/7 1/7 1/7 1/7];
>> E=X*p';
>> x2=X.*X;
\begin{align}
E[X]&=4\\ 
E^2[X]&=20\\
V[X]&=4
#+END_latex


But not the last 2 lines of the original matlab code.

Anybody with a better  know of the matlab-shell can explain that to me
please? In order to test the code I should send a bunch of more
functions, which I attach for convince.


Regards

Uwe Brauer

[Matlab-emacs-discuss] bug in matlab-shell-run-cell (comments)

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

From: Uwe Brauer <ou...@ma...> Subject:  To: 
matlab-emacs-discuss <mat...@li...> 
Date: Mon, 26 Feb 2018 10:19:21 +0100 (2 days, 4 hours, 51 minutes 
ago)  Hi  Please look at the following example attached, the 
function matlab-shell-run-cell exectued the code which is 
sourrounded by %{ disp('Code') %}  While according to the 
documentation, this is a valid commented out region.  Matlab 
internal editor respects the comment and does not execute it.  So 
I presume we do not really support that sort of comments, do I 
have a deja-vu here? Did we discuss this already? Any comments?  I 
just tested matlab-shell-run-region also execute the commented 
region.  So the problem is  deeper, then just the cell function. 
Uwe Brauer 

Re: [Matlab-emacs-discuss] indention stopped working?

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

>>> "Eric" == Eric Ludlam <Eri...@ma...> writes:

   > John Ciolfi and I looked at your example. You are missing a close ].
   > See the arrow I added below to your code snippet.

Ah. You are right. That was the file with the mistake. I am really sorry
and feel stupid.

   > I'm thinking that adding highlighting to keywords inside a [ ] block
   > might be a useful feature to make this more clear. Not sure how much
   > that would slow things down as detecting that is probably expensive.

That would be helpful, but I can't judge the speed reduction. BTW I
complained in another mail that 2016b and above is really slow on my
machine. That is only true, if I start the whole beast. For emacs the
change and the startup via no desktop, it is not notable.

Sorry again..

Uwe 

Re: [Matlab-emacs-discuss] indention stopped working?

[Eric L. <Eri...@ma...>]

John Ciolfi and I looked at your example.  You are missing a close ].   See the arrow I added below to your code snippet.

I'm thinking that adding highlighting to keywords inside a [ ] block might be a useful feature to make this more clear.  Not sure how much that would slow things down as detecting that is probably expensive.

Eric

-----Original Message-----
From: Uwe Brauer [mailto:ou...@ma...] 
Sent: Monday, February 05, 2018 4:23 PM
To: Eric Ludlam <Eri...@ma...>
Cc: Uwe Brauer <ou...@ma...>; matlab-emacs-discuss <mat...@li...>
Subject: Re: [Matlab-emacs-discuss] indention stopped working?

>>> "Eric" == Eric Ludlam <Eri...@ma...> writes:

Hi Eric

Thanks 
> Your examples work for me , but I don't have Emacs 27.

I tried out Emacs 26, same problem, I deactivated any auto-fill function.

> The only thing I can think of is some sort of auto-fill behavior is 
> going off, but that seems unlikely with just an indent-region.


What is strange is the following

If the file just contains

for k=1:N+1
x(k+1)=x(k)+h;
y(k+1)=f(x(k+1));
z(k+1)=g(x(k+1));
end

Then indent-region does the right thing

But if the file has
f=@(x)[x*exp(-x)];
g=@(x)[exp(-x^2);  <----------------
for k=1:N+1
x(k+1)=x(k)+h;
y(k+1)=f(x(k+1));
z(k+1)=g(x(k+1));
end

Then the indent-region function acts in a bizarre way. Can somebody please confirm this?

Re: [Matlab-emacs-discuss] indention stopped working?

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

>>> "Eric" == Eric Ludlam <Eri...@ma...> writes:

Hi Eric

Thanks 
> Your examples work for me , but I don't have Emacs 27.

I tried out Emacs 26, same problem, I deactivated any auto-fill
function.

> The only thing I can think of is some sort of auto-fill behavior is
> going off, but that seems unlikely with just an indent-region.


What is strange is the following

If the file just contains

for k=1:N+1
x(k+1)=x(k)+h;
y(k+1)=f(x(k+1));
z(k+1)=g(x(k+1));
end

Then indent-region does the right thing

But if the file has
f=@(x)[x*exp(-x)];
g=@(x)[exp(-x^2);
for k=1:N+1
x(k+1)=x(k)+h;
y(k+1)=f(x(k+1));
z(k+1)=g(x(k+1));
end

Then the indent-region function acts in a bizarre way. Can somebody
please confirm this?

Re: [Matlab-emacs-discuss] indention stopped working?

[Eric L. <Eri...@ma...>]

Your examples work for me , but I don't have Emacs 27.

The only thing I can think of is some sort of auto-fill behavior is going off, but that seems unlikely with just an indent-region.

Eric

-----Original Message-----
From: Uwe Brauer [mailto:ou...@ma...] 
Sent: Saturday, February 03, 2018 10:29 AM
To: matlab-emacs-discuss <mat...@li...>
Subject: [Matlab-emacs-discuss] indention stopped working?

Hi

I swore this worked till recently:


I have 

k=1:N+1 x(k+1)=x(k)+h; y(k+1)=f(x(k+1)); z(k+1)=g(x(k+1)); end 
 


So I mark it and run indent-region

And I receive

 
       for k=1:N+1 
    x(k+1)=x(k)+h; y(k+1)=f(x(k+1)); z(k+1)=g(x(k+1)); 
       end 


Which is absurd, I should have obtained


for k=1:N+1 
    x(k+1)=x(k)+h; y(k+1)=f(x(k+1)); z(k+1)=g(x(k+1)); end  Now I run matlab-indent-sexp

And obtain again

       for k=1:N+1 
    x(k+1)=x(k)+h; y(k+1)=f(x(k+1)); z(k+1)=g(x(k+1)); 
       end 

What is up here?

Or is the culprit GNU emacs 27? 


Uwe Brauer 

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Matlab-emacs-discuss mailing list
Mat...@li...
https://lists.sourceforge.net/lists/listinfo/matlab-emacs-discuss

[Matlab-emacs-discuss] indention stopped working?

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

Hi

I swore this worked till recently:


I have 

k=1:N+1 x(k+1)=x(k)+h; y(k+1)=f(x(k+1)); z(k+1)=g(x(k+1)); end 
 


So I mark it and run indent-region

And I receive

 
       for k=1:N+1 
    x(k+1)=x(k)+h; y(k+1)=f(x(k+1)); z(k+1)=g(x(k+1)); 
       end 


Which is absurd, I should have obtained


for k=1:N+1 
    x(k+1)=x(k)+h; y(k+1)=f(x(k+1)); z(k+1)=g(x(k+1)); 
end  Now I run matlab-indent-sexp

And obtain again

       for k=1:N+1 
    x(k+1)=x(k)+h; y(k+1)=f(x(k+1)); z(k+1)=g(x(k+1)); 
       end 

What is up here?

Or is the culprit GNU emacs 27? 


Uwe Brauer