Re: [Matlab-emacs-discuss] Call for testers, code not in master (yet)

[Karthik C. <kar...@gm...>]

Update: This bug goes away if I change my m file encoding from utf-8-unix to us-ascii-unix. Matlab's fileread() routine only seems to works correctly if the m-file is encoded as ASCII (and maybe a few other non-UTF8 systems, I'm not sure.) This exception should probably be handled by the code in the function emacsrunregion.

Karthik

Karthik Chikmagalur <kar...@gm...> writes:

> - What is the best way to report bugs pertaining to the modernize
> branch?
>
> - Bug in emacsrunregion.m:
>
> There seems to be a difference in the way Emacs and Matlab count
> characters in my m files. The region I send from Emacs and the region
> being used in the line
>
> ,----
> | ETXT = TXT(startchar:endchar);
> `----
>
> in `emacsrunregion.m' are often (but not always) off by a characters,
> leading to inconsistent behavior and errors.
>
> For example, with this code I wrote in `test.m':
> ,----
> | %% Load mat file with white noise input
> | 
> | load 'white_noise_input.mat'
> | % 
> | % % Use same input for discretized only and discretized+lifted
> | % timesteps_lifted = 16384
> | %% Initialize the lifted system
> | 
> | N = 100;
> | 
> | % Damping coefficient
> | % -------------------
> | damping_coeff = 0.000; % no damping
> | % k = 0.005; %damping coefficient, was 0.005 -> damping ratio = k/(1*omega) = 0.025
> `----
>
> I select the following region and run `matlab-shell-run-region':
>
> ,----
> | %% Initialize the lifted system
> | 
> | N = 100;
> | 
> | % Damping coefficient
> | % -------------------
> | damping_coeff = 0.000; % no damping
> `----
>
> which prints
> ,----
> | >> emacsrunregion('./test.m',164,287)
> `----
>
> This throws a bunch of errors. Here is the region that's actually being
> evaluated (note the same character range):
>
> ,----
> | >> TXT = fileread('./test.m');
> | >> TXT(164:287)
> | 
> | ans =
> | 
> |     '6384
> |      %% Initialize the lifted system
> | 
> |      N = 100;
> | 
> |      % Damping coefficient
> |      % -------------------
> |      damping_coeff = 0.000; % '
> `----
>
> Which is off by a few characters. All my m files are encoded as
> utf-8-unix, if that helps.
>
> - Also, I had a question regarding how `eval' works in Matlab. It
>   appears that it does not maintain state, i.e., eval sets up an
>   environment for execution that it discards once it's done executing
>   the region/line that was sent. This makes interactive Matlab coding
>   (switching back and forth between the m file and `matlab-shell')
>   almost impossible, where in earlier versions the state in
>   `matlab-shell' was persistent. Is this intended behavior?
>
> Karthik Chikmagalur
>
> Uwe Brauer <ou...@ma...> writes:
>
>> Hello everybody
>>
>> * New Branch: modernize 
>>
>> Currently Eric is submitting a lot of changes to the emacs matlab
>> packages. Thanks a lot Eric!
>>
>> Some of the changes/feature which are currently under work have been
>> announced by Eric in a mail to the list, sent at 11 of October.
>>
>> His contributions are committed not to master but to a new branch,
>> called called modernize, so it is not merged into master, yet.
>>
>> I am trying to test his commits, but I would prefer if more users would
>> join me to get a better feedback.
>>
>> Here is what you have to do.
>>
>> * Git users
>>
>> git clone https://git.code.sf.net/p/matlab-emacs/src matlab-emacs-src
>>
>>  git checkout modernize
>>
>> * Mercurial users
>>
>>  install and activate the hg-git plugin
>>
>>  hg clone https://bitbucket.org/durin42/hg-git/
>>  
>>  then in your .hgrc file something like this
>>
>>  hggit = /home/user/src/hg-git/hggit
>>
>> [git]
>> export-additional-refs.named-branch-heads:revset   = head() and not closed()
>> export-additional-refs.named-branch-heads:template = refs/heads/hg_named_branch/{branch}
>>
>>  hg clone https://git.code.sf.net/p/matlab-emacs/src emacs-hg
>>
>>
>>  hg up modernize
>>
>>
>> * Compilation 
>> Run 
>>
>>  make
>>
>> * Set the path
>>
>> Now set the path in your init file, for example 
>>
>> (setq load-path (cons (expand-file-name "~/emacs/site-lisp/packages/src/Matlab/matlab-emacs-src")  load-path))
>>
>> Add 
>>
>> (load-library "matlab-load")
>> (matlab-cedet-setup)
>>
>> Optionality 
>> (semantic-mode 1)
>> (global-semantic-stickyfunc-mode 1)
>> (global-semantic-decoration-mode 1) 
>>
>>
>> * News
>>
>>     1. The debugger works better now, at least this is my impression.
>>
>>     2. matlab-shell-run-cell has been completely rewritten. Now it
>>        possible to have a cell which execute a function, which is
>>        defined at the end of this very same file.
>>
>>     3. Faster load time by moving rarely used content into other files.
>>
>>     4. Support for some older Emacs' has been deleted.
>>
>>     5. Clicking on errors or using next-error in matlab-shell should work better.
>>
>>     6. tab completion in the shell was revamped (but should act mostly
>>        the same)
>>
>>     7. Other bug fixes to editing ML code.
>>
>>
>> Regards
>>
>> Uwe 
>> _______________________________________________
>> Matlab-emacs-discuss mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matlab-emacs-discuss