Ed Leafe wrote: > On Dec 2, 2008, at 6:34 PM, Paul McNett wrote: > >> I still think it is correct to save only overridden/changed >> properties, even for >> classes. Think of the equivalent in pure-python code. You don't see: > > I started out from that position, and quickly found that things got > real messy real fast with more than one level of inheritance. While > I'd like to be able to say "Don't subclass more than one level", > that's an arbitrary limitation.
Definitely, and seems like a red herring. We should make it correct, no matter what it takes. > Take an example with 3 levels of inheritance: to find out what the > default value of a property should be for an instance, I have to find > its superclass and see what it is there. > If it's not defined, I then > have to find the super-superclass and check that. If it's not defined > there, then I have to check the super-super-superclass. Only if it's > not defined there can I assume that the instance should use the default. The defaults will be the default from the single superclass. You don't have to ask layers of superclasses. Instantiate your subclass, and then find out what the property value is. That is the default. > The only practical solution was to store all properties on classes, > and only changed properties on instances. Seems like an implementation problem? Why not instantiate a one-off instance, get the defaults, and go? Paul _______________________________________________ Post Messages to: [email protected] Subscription Maintenance: http://leafe.com/mailman/listinfo/dabo-users Searchable Archives: http://leafe.com/archives/search/dabo-users This message: http://leafe.com/archives/byMID/[EMAIL PROTECTED]
