[kplato] Application Class Structure

[Thomas Zander]

On Sun, Jun 24, 2001 at 09:36:35PM +0100, John D Lamb wrote:
> Thomas Zander wrote:
>  
> > I am mainly a Java programmer, sorry for the default question; why are
> > methods which are implemented in parent classes not displayed in the child
> > classes?
> > Makes it hard to figure out functionaility.
> >
> I think this is a feature of doxygen. There is a "List of all members",
> that should help. I think it's six and half-a-dozen whether this is
> easier than javadoc. It's harder in javadoc to identify which class
> implements a member function.

hmm, ok.

> > I honestly admin that the risk/duration discussion was over my head, but
> > the implementation of the tree of classes seems a bit to much for me.
> > I think one class should suffice, maybe have multiple constructors..
> > The main reason to use one class is becuase the basic duration is a data
> > object, all the classes you have under them are calculations on that object.
> > I suggest to move those to the duration class (make it know things about
> > itself).
> > Maybe add a number of static methods which return a duration object..
> I'm not yet convinced of this, but its good to think about it. 

I forgot one argument, if you create one class you can just execute a method
to alter the data on that object, in your solve you should delete the object
and create the correct object in its place. 
This means that the controlling code needs to know the contents of the risk
/duration object.
This defies the OO approuch, where the object itself is responsible for its
data, not the controlling code.

-- 
Thomas Zander                                            zander@earthling.net
The only thing worse than failure is the fear of trying something new