>If you were stringent you wouldn't instanciate this abstract class but only
>subclassed concrete classes for concrete devices. Abstract does not
>have anything to do with not accessing (meta) data.
>
>In fact in an earlier discussion we found that VFP is good in a kind
>of abstract classes thet get concrete at runtime and that this isn't
>a bad decision.
>
>But like Christof said it's hard to understand why a device should
>suddenly change. If you interpret your concrete device object as that
>device and have a seperate communication class which instanciates
>the proper device class, then you can split functionality.
>
>Why is that data access your concern? Because you already
>programmed the initial configuration of the device object is taking
>place from outside? Or because all data access is done by data
>access classes and you want to make the device class work
>with those data access classes on the one side, to be stringent,
>but you also want it to be a self contained class which does not
>only work in conjunction with this kind of data acess object?
>
>Bye, Olaf.

I realized yesterday that I was going about it in the wrong way. I now know that a change of the device should be initiated from outside the device class.

Thank you!