From graphics-request at octave dot org Fri Mar  4 10:05:09 2005
Subject: Re: Proposal for a GUI extension to Octave
From: Driss Ghaddab <driss dot ghaddab at free dot fr>
To: Muthiah Annamalai <ec10130 at nitt dot edu>
Cc: graphics at octave dot org
Date: Fri, 04 Mar 2005 17:11:32 +0100

Hi Muthiah,

[Incidentally, cmo 0.0.3 is released and has GCC bug fixes, more error 
messages. Timer module has race-conditions removal and new implemented 
AveragePeriod and InstantPeriod]

Muthiah Annamalai a écrit :
> Copying functionality is perfectly allowed under copyrights,
> as long as your work is original. [no code copying].
> Patents only abuse[or in patent holders terms protect] software 
> functionality.

So I need to do some patent search.

> As for implementing a GUI, I would like you to choose compile
> time widget selection, and choose writing wrappers over
> generic toolkits [GTK,Qt,Win32,Fltk,WxWindows, etc]. to implement
> what our neighbour (Matlab) widgets like
> [Buttons,Labels,Text,PopupMenus, Radioitem, Slider, etc].

In fact, I'm thinking about creating an abstract class that does all 
widget independent work. It'll provide virtual functions that do the 
real widget related work. As all widget sets use an event driven 
paradigm, I can factorize many things in the base class. I'll just need 
to check if MS Windows UI and Cocoa are also event driven.

> As far as runtime widget selection is involved you may easily
> add / remove toolkits at users option. But you may aso specify the default 
> toolkit in a .octaverc file or something. That mustnt stop us form
> using variety of free software to our advantage. ;-)

So there is a demand for runtime selection of widget set...

> We [Octave-GTK team] started with an abstract widget hierarchy to
> make a compile time binding to Matlab like API for Octave GUI.
> But it didnot really take off.

> As a team lead of Octave-GTK project, I can assure of a few things 
> we are currently having in mind.
> 
> 1: Octave-GTK team is porting Glade GUI builder for working with
> Octave, so that it can be the GUIDE replacement.

That would be great. I may try to help you there.

> 2: We are looking into reverse engineering '.fig' format, and writing a
> glade converter tool from '.fig' and vice-versa.

> 3: Octave-GTK team also has plans for Matlab API implementation, but
> we are not moving as fast as we would like to.

As long as we use matlab like syntax for the scripts, we can join forces 
for glade (or qtdesigner, or...)...

Cheers,

Driss Ghaddab