Re: [Figleaf-developer] Current code changes

[Greg S. <ste...@on...>]

I think having the capability to modify behavior on a class-by-class 
basis is a good thing. There could be a good reason for someone to alter 
their implementation of Observable for a particular class. Since we 
can't foresee all possible usage scenarios, I wouldn't want to take that 
capability away from the developer. Of course with power comes 
responsibility: if they do it, they need to understand the consequences 
of how that object will behave in the application stack.

I intentionally left the interface Informative without any methods of 
its own. It's basically a tag interface for knowing an object implements 
all three of Observable, ClassDescriptor, and Validatable (and perhaps 
we will add more later). My thought was if a class (or its helper) 
implements ClassDescriptor and Validatable, then the Introspector would 
detect that and provide a default implementation of Observable for it, 
thereby creating a proxy that implemented Informative. So the class (or 
its helper, metadata, etc) could provide implementations for none, some, 
or all of the sub-interfaces of Informative, and the 
Introspector/ProxyFactory would do the right thing to make a proxy 
instance that implements Informative completely.

Greg


sam...@ma... wrote:

>Do we really want that? My code does that for the developer - the UI will depend
>on these events being properly fired in order to maintain system integrity, as
>such I think its something that should be under our control.
>
>Quoting Greg Steckman <ste...@on...>:
>
>  
>
>>Note that Informative extends Observable. So if you implement 
>>Informative, then it means you must also implement Observable.
>>
>>Also, not all methods will be called via Property objects. Consider 
>>those objects invoking methods directly on the object (that is, those 
>>that know nothing about the Informative interface).
>>
>>Greg
>>    
>>
>
>
>
>sam
>http://www.magpiebrain.com/
>
>  
>