Re: GUI design

[Daniel J Sebald]

On 03/23/2012 06:39 PM, J. Luis wrote:
> On 23-03-2012 22:55, Clark Dunson wrote:
> >  Mathworks has offered GUI building functionality as part of the product for a 
> > while.  If we be Qt, then how to create GUI building in the m-file?  Focus by 
> > the core group on devising the GUI/API layer that allows us all to create GUI 
> > might be time very well spent.
> >
> >  Of course this is probably already talked about, perhaps even already written 
> > and available.
>
> Hi,
>
> I have tried to be keep out of this discussion as most participants sea to have
> their minds set but this touched a subject dear to me ... and so I go
>
> I'm a ML as well as CLI relatively experienced user and I find it unthinkable to
> use only the CLI when I can use the IDE. Did you guys never used breakpoints in
> the ML editor? How can you live without it once you did it once? CLI???????????

Well, my take on this is that breakpoints can be helpful.  (Keep in 
mind, pertaining to the original question, I'm assuming GUI Octave 
doesn't have breakpoints or may not be good at it.)  In writing compiled 
code (i.e., C, C++) a debugger and breakpoints are wonderful.  But 
Octave scripts have such a high level of memory/data control and are so 
descriptive in terms of error messages that I don't think breakpoints 
are that necessary, at least not for me.  That being said, in the GUI 
development portion of some IDE, breakpoints might be helpful.  It might 
be nice to break out of that to determine, say, what particular hunk of 
script is run by pressing a button and so on.


> A second point is that I also find it unthinkable to try force students to use a
> command window and hide from them that Matlab exists.

You mean the IDE exists, I assume.


>  CLI only defenders, just
> try to do that and please report your experiences with us.

Well, when John originally wrote Octave the idea was to assist students 
in analyzing data.  Students haven't changed that much.


> but, no roses on the way. Octave has lots of bottlenecks and an issue even more
> important than IDE is Octave performance
>
> (I'm not going to make friends here). While Octave shows up in benchmarks
> positions like in http://julialang.org/ I don't foresee a bright future ahead.
> And please do not invocate users guilty on not writing fully vectorized code.

Yes, I agree.  I think in the past week or two we've identified some 
performance issues.  I don't believe the fixes are too difficult though 
(not simple, but not a major undertaking).  The code is organized well 
enough that some tweaking and debugging might do.

Dan