🚀 go-pugleaf

Geoffrey:

My take is that models should be as separate and distinct from their
views as possible.  That distance is key, in my mind, to good MV
architecture.

Views tend to have some visual aspect that works in Designer tools but
you don't just drop down a non-trivial model and get feedback.  If you
happen to have a fully constructed model around you can bind a view to
it in a visual environment, of course.  To me, it doesn't make sense
to burden the model (a container for data) with all the overhead to
make it "better in a bean/visual environment".  You should be able to
pass a model around efficiently, just like any other data structure. 
I don't think it's a safe assumption that a model will be bound to a
visual-widget always.  The TreeModel in JFC is pretty handy all by
itself, for example.  If you want to "see" it, then you bind it to a
view object that can do something with it (JTree, but it doesn't have
to be that obvious").

I also like the factory concept you see more of in JFC.  There is a
model factory that you can go get a base model or some variant from at
the point you need it (which could be runtime and not designtime).

Just my .02., of course.

-Rich

Geoffrey L. Goetz wrote:
> 
> "Ross M. Dukeshier" wrote:
> > Just an aside - if I remember correctly, one of the requirements to
> > being a java bean was that the class implement a parameterless
> > constructor.  You referenced DefaultTreeModel - it doesn't have one...
> >
> > Didn't check if that was true of all the models, but I figured it may
> > explain at least one...
> >
> 
> As I feared, my original questions and attempts to discuss have been
> misinterpreted.  I was not trying to ask "why doesn't the square peg fit
> into the round hole" rather I was trying to discuss the reasons "why the
> square peg wasn't a round peg to begin with?".  I will shoulder the
> blame of the misinterpretation.
> 
> Should the models behind the visual components be developed and extended
> separately as JavaBeans and if so, would easier access to these models
> be of benefit to developers?  Is keeping the model separate from the
> visual component in the IDE a good or bad thing?  Are people defining
> custom models similar in responsibility to DataExpress's DataSet
> Objects?  I feel that the visual components themselves have been and
> will continue to be unnecessarily overburdened to also include an
> interface into the model's attributes and events.  Should the details
> MVC in swing survive and propagate into the day to day development, or
> is it just a good idea that is best kept under the hood so to speak?
> 
> IMHO, developing a rich set of models that can easily be added to an
> application that are not tied to a database would be of considerable
> value.
> 
> --
>                            \\|//
>                          | (o o) |
>   =-=-=-=-=-=-=-=-=-=-=-=-=-(_)-=-=-=-=-=-=-=-=-=-=-=-=-=
>  _|__         ______                                  __|_
> (____)       / _____)  ___   ______  _____  _____    (____)
> (____)      / /  ___ / ___) / __  / / ____)/ ____)   (____)
> (____)     / /___/ // ___) / /_/ / / ___) / ___)     (____)
> (__)      /_______//_____)/_____/ /_/    /_/           (__)
>   |                                                     |
>   |        http://home.columbus.rr.com/ggeoffre         |
>   |                                                     |
>   |                    .oo0O   O0oo.                    |
>   |                    (   )   (   )                    |
>   =-=-=-=-=-=-=-=-=-=-=-\ (-=-=-) /-=-=-=-=-=-=-=-=-=-=-=
>                          \_)   (_/

-- 

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Rich Wilkman
Senior Quality Assurance Engineer
Borland JBuilder Team

If you see something that looks like an opinion up there,
it's mine and mine alone.

Mail sent to me from the newsgroups won't bounce but will
be ignored.  If I have time, I'll respond, but you  have
a better chance if you confine your posts to the newsgroups
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=