matlab-emacs-discuss

[Matlab-emacs-discuss] Patch for matlab.el regarding command completion in MATLAB shell

[David E. <de...@ra...>]

Attached to this mail is a patch for matlab.el which improves command
completion in the MATLAB shell. In the current CVS version of matlab.el,
comint filename completion is done when the cursor is in an imcomplete
string. This is nice for commands like 'addpath', but is wrong for
completing things like

set(gca,'Cam <TAB>

After applying the attached patch, the above completion should work as
expected. It also works for completing commands and variable names
inside commands like "plot(...)" and others.  Additionally, it includes
a small change in 'matlab-shell-completion-list', which disables
scrolling in the shell buffer when retrieving the completions.

The patch should be applied against the current CVS version of
matlab.el.

Regards,
David

Re: [Matlab-emacs-discuss] Patch for matlab.el regarding command completion in MATLAB shell

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

Thanks for the patch.  I merged it into my local copy, and have read it over.  I haven't tried it yet, but the differences look good.

Hopefully, I will test it out, and check something in tomorrow.

Thanks
Eric

-----Original Message-----
From: mat...@li... [mailto:mat...@li...] On Behalf Of David Engster
Sent: Thursday, July 31, 2008 4:37 PM
To: mat...@li...
Subject: [Matlab-emacs-discuss] Patch for matlab.el regarding command completion in MATLAB shell

Attached to this mail is a patch for matlab.el which improves command
completion in the MATLAB shell. In the current CVS version of matlab.el,
comint filename completion is done when the cursor is in an imcomplete
string. This is nice for commands like 'addpath', but is wrong for
completing things like

set(gca,'Cam <TAB>

After applying the attached patch, the above completion should work as
expected. It also works for completing commands and variable names
inside commands like "plot(...)" and others.  Additionally, it includes
a small change in 'matlab-shell-completion-list', which disables
scrolling in the shell buffer when retrieving the completions.

The patch should be applied against the current CVS version of
matlab.el.

Regards,
David

Re: [Matlab-emacs-discuss] Patch for matlab.el regarding command completion in MATLAB shell

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

Hi,

  Your patch worked well for me in my quick tests.  I checked in this
change, and also flipped the default value of the variable
`matlab-shell-ask-MATLAB-for-completions'.  This was originally nil because it wasn't very stable.

Thanks for this great improvement.
Eric

-----Original Message-----
From: mat...@li... [mailto:mat...@li...] On Behalf Of David Engster
Sent: Thursday, July 31, 2008 4:37 PM
To: mat...@li...
Subject: [Matlab-emacs-discuss] Patch for matlab.el regarding command completion in MATLAB shell

Attached to this mail is a patch for matlab.el which improves command
completion in the MATLAB shell. In the current CVS version of matlab.el,
comint filename completion is done when the cursor is in an imcomplete
string. This is nice for commands like 'addpath', but is wrong for
completing things like

set(gca,'Cam <TAB>

After applying the attached patch, the above completion should work as
expected. It also works for completing commands and variable names
inside commands like "plot(...)" and others.  Additionally, it includes
a small change in 'matlab-shell-completion-list', which disables
scrolling in the shell buffer when retrieving the completions.

The patch should be applied against the current CVS version of
matlab.el.

Regards,
David

Re: [Matlab-emacs-discuss] Patch for matlab.el regarding command completion in MATLAB shell

[David E. <de...@ra...>]

Eric Ludlam <Eri...@ma...> writes:
> and also flipped the default value of the variable
> `matlab-shell-ask-MATLAB-for-completions'.  This was originally nil
> because it wasn't very stable.

Well, there's still one thing that is a bit annoying: after working with
the MATLAB-shell in Emacs, the history is cluttered with those
'matlabMCRprocess' commands. Is there some way to avoid this? Maybe some
other MCRprocess-command which temporarily disables the history? (I
guess those are for internal use and not documented anywhere publicly?)

-David