Re: The death of WIDED (was: Interactively building GUIs in IDL 5.1?)

["Richard G. French" <rfrench(at)wellesley.edu>]

David Fanning wrote:
> 
>
> If I were to tell you that I had a piece of software that
> would allow you to build widget programs interactively
> (I mean by this that you could drag and drop "widgets"
> onto a palette, set properties by double clicking the
> widgets, add event handler code by means of a built-in
> editor, be able to recover and re-edit the IDL code it
> generated, etc.), what would you be willing to pay for it?
> 

I think this is a useful topic for discussion. As has been pointed out
before in this news group, there is a whole range of software, from
the clever 1-line routine that determines the ENDIAN nature of the
particular machine, to libraries of routine available on the web for
medical or astronomical data analysis, for example, to commercial
products like ENVI (and PV-WAVE?) that have IDL as their basis but
provide professional-quality enhanced features. 

One of the great things about using IDL is the amount of free
advice and help that is available on this newsgroup. Everything
is 'freeware' so far - I have not seen IDL shareware around. 
Also, to remain competitive, all computational software needs
to add new features on a regular basis. So, my first reaction to 
David's question was that RSI should provide this capability for free,
but as I think about it, I realize that there are other models
worth considering. For example, with MATLAB, you can buy the
basic program, but add functionality with add-on 'toolboxes'.
I have found these toolboxes pretty expensive, but well-documented,
very useful, and supported by the makers of MATLAB. I am willing
to pay the price for something that works and meets professional
standards. Home-brew software nearly always fails that test in some way
or another.

I have tried twice to create a large widget-based program, and I
have had to relearn the tricks of the trade each time, due partly
to the many enhancements in the way RSI has set up widgets. There
are now approximately 580,358 keywords available for each widget
program, and some of us would like to avoid having to learn about
each and every one of them just to put together a widget. I know
that it would increase my produtivity enormously if I could build
widgets with a widget-builder. When I build WEB pages, I do it by
hand, without a composer program, but I shamelessly copy examples
from every nice web page that I see. Widgets are a lot tougher than
HTML, and I have gone to David Fanning's book and web page quite a bit
to try to learn a proven style for widgets that avoid the dreaded
XMANAGER error messages.

I'd be willing to pay someone for a widget-building program as
David describes. Although it would be nice to think that I could
see the source code for the program, I don't ask that of the Mac
and PC programs I buy, nor of ENVI, and a compiled savefile would
be fine by me. In MATLAB, there is a licensing scheme by which the
toolboxes are added to the license, and I would have no objection
to this kind of restriction so that the software provider got paid for
each instance of the program used. I think this kind of protection is
essential for the programmer, or you end up with a million pirate
copies and no financial return for the programmer.

How much would I be willing to pay? Tougher question, but certainly
as much as $150 and probably not more than $250.

Other opinions welcome, even if they disagree with mine! I'd love to see 
such a program developed.

Before closing, how about a package of object oriented programs
that are complete and do useful simple things? There is a steep
learning curve for them, and I'd be willing to pay for them as well,
although I think that RSI should definitely work on supplying more of
them themselves. I know that they are not doing much to help the
neophyte object-oreinted programmer like me take the giant step to
trying to build a large OOP application.

Dick French
Astronomy Dept
Wellesley College