[Opensource] Future Code for Expresso 5.1 and beyond [LONG THINK TANK ARTICLE]

[Michael Rimov]

At 05:52 AM 11/4/2002 -0800, you wrote:
>Choice 1 and 3 may have similar code bases, but would it not be better
>encapsulation and possibly better re-use to go with choice 1?
>
>Just playing devil's advocate,
>David Lloyd
>
> > >        DBObject.getField().asString()/.asDate()/.asInteger()  etc.
> > >
> > >                or
> > >        (Date)DBObject.getField()/(String)DBObject.getField() etc.
> > >                or
> > >
> > >        DBObject.getFieldString()/.getFieldInt()/.getFieldDecimal()

Choice #1 DOES give us the ability to have custom Field types.  For 
example, we can have a specialized DataField type called GIFField that 
contains all the gif metadata.  Specially formatted fields are also a 
possibility.  If we use an XML configuration, we could specify the class to 
use for the field and then give us another layer of abstraction.  Thanks 
for playing opposite advocate here in the sense that it gives me a chance 
to better think out things :)  I've used #1 in C++ Builder with good 
results... but, like Larry said, it IS more typing, and it's a matter of 
whether or not the migration is worth it.

I suppose if people started contributing custom data fields, then it 
suddenly would become worthwhile to go with #1, but if nobody is interested 
then it's not worth it.

                                                 -Mike