| Re: Widget Objects [message #19143] |
Mon, 28 February 2000 00:00 |
Mirko Vukovic
Messages: 124
Registered: January 1996
|
Senior Member |
|
|
In article
<hamillnumerics-2502002135370001@user-37kafqa.dialup.mindspring.com>,
hamillnumerics@mindspring.com (Jim Hamill) wrote:
> An object-oriented approach to IDL widgets works nicely. Readers of
this
> newsgroup who'd like to see what I'm talking about are invited to
check
> out ...
>
>
http://hamillnumerics.home.mindspring.com/hamillnumerics/idl /javaconcept
s/widgetobject/widgetobjects1.html
>
> Has someone already done this?
>
My recent thoughts on the subject stem from the fact that I really
dislike building widgets, and from a need to build a GUI for a rather
complex object.
Now, as luck would have it, in this complex object, I had numerous lists
of attributes, for lack of a better name (not properties). While
objects have properties that are basically IDL variables, the attributes
are higher level stuff, stuff with meaning: Flags, range specifiers,
color table indices, one-of type variables (var can be one of A,B,C
say), variable scans (lin, log, exp, power, name it). These attributes
are all objects.
One advantage of attributes, is that they can check their own validity,
they can have their own documentation! This was actually the initial
motivator: no need to write documentation, _include_ it in the code.
The code could then print out the keywords necessary to set the
attributes, the constraints on the attributes, the works.
The attributes can also build their own GUI's on the fly. There is only
one or two ways to build a GUI to set a clear a flag after all.
Building a GUI then invloves setting a base widget, going through a list
of attributes and building it up.
My thoughts have evolved somewhat. Now I think of a GUI as a UI to a
routine or an object. There should be very little logic in it. The
logic should all be in the IDL routine or object. A GUI is composed of
groups (which themselves can contain elements of this self-referential
sentece), or data fields (by that I mean higher level stuff, like
flags, ranges, etc), user action initiators (run routine X), and GUI
state controllers (like more or less detail) and placeholders (I'm not
sure about them yet, but they would be for objects that can change. For
example an IDL function can accept a scalar or a vector).
I've been influence to a large extent by the LaTeX typesetting system in
this regard. The GUI layout would be specified logically, and would
deal mainly with the look of the thing. The appearance of individual
elements would be controlled by the object that corresponds to its
GUI field. It seems to me that this eases changes to the layout
considerably.
Each of these object display properties that can be changed on the fly.
It would then become possible to dynamically change the look of a GUI.
One very important thing to me would be that GUI components would be all
parts of a large inheritance tree. I should be able to change the
background color, or the base font size of the base GUI, and the changes
to affect all the children. It would seem to me that building GUI's
across platforms would be much easier, since one would need to change
the paramaters at only one place.
One could make a group self-standing on the fly, so it is in a widget by
itself, but still part of the whole logical group, and the
re-incorporate it again into the base widget.
Caveat? None of the GUI stuff has been done yet! Mostly thinking and
pondering, writing specs, doodling on paper, but no hard coding.
Something of this size, I would approach carefully, after a whole lot
more of doodling on paper.
Un-answered questions? Well, I'm not sure about the purpose of
event-bubbling up in the widget world (when event_fun returns an event).
I have not run across an application that needed it. I think that in
the scheme of things laid out above, such a feature would not be needed.
Since GUI is just a front end, if a result of the user action processing
(first event) needs further processing, the underlying programs should
take care of that.
I have not really thought hard about graphics and cursor actions in this
scheme. And there are probably a few other issues that I can't think of
right now.
Mirko
Sent via Deja.com http://www.deja.com/
Before you buy.
|
|
|
|
| Re: Widget Objects [message #19149 is a reply to message #19143] |
Fri, 25 February 2000 00:00  |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
Jim Hamill (hamillnumerics@mindspring.com) writes:
> An object-oriented approach to IDL widgets works nicely. Readers of this
> newsgroup who'd like to see what I'm talking about are invited to check
> out ...
>
> Has someone already done this?
Uh, no. Never occurred to me.
Cheers,
David
P.S. Excuse me while I go bang my head against the wall here. :-(
--
David Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438 E-Mail: davidf@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
