On 6/25/11 13:59 , Niclas Hedhman wrote:
Meaning that a "PojoMixin" could deal with mapping getters/setters style properties to internal state. This would lower the Qi4j impact in sensitive projects which have a 'low sophistication level' as the rule of thumb. IF we could also manage to do;public class MyValue { private String value; public void setValue( String value ) { this.value = value; } public String getValue() { return value; } } and get that to be 'State' in a generic fashion, and we would get a huge boost in applicability.
If you want that to be the whole definition of a Value, it won't work. It still needs to be a mixin with an interface. The idea was that we don't have to require that declared types extends ValueComposite/EntityComposite/TransientComposite, *but* the final proxy will, anyway, so casting to "ValueComposite" will go through, as well as instanceof checks. That being said, we could probably simplify it down to "Value" (extends Composite), "Entity" (extends Composite and Identity), and "Transient" (extends Composite), but with no methods, i.e. they are purely marker interfaces for identification purposes.
Nothing other than the occasional type check in module.entities() where it happens I stick things in wrong places, but that is not compelling enough.
See above. The marker interfaces we could add anyway. /Rickard _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
