Re: Cocoa et al as HCI usability problem

FROM : Peter Duniho
DATE : Mon May 19 23:55:08 2008

On May 19, 2008, at 2:20 PM, Jean-Daniel Dupas wrote:

> And as we are here, note also that Key-Value-Coding uses dynamic 
> properties of the language.


Yes, it does.

> OK, implementing valueForKey: and setValue:forKey: is probably easy 
> using introspection.


Likewise reflection.  And in .NET, the same kind of binding 
functionality is fully supported.  The basic version uses something 
like KVC, but with reflection to grab the property and associated 
event.  I don't use .NET binding very much, but I know there's also a 
more strongly-typed version of binding that doesn't rely on naming 
conventions or reflection.

> So now, have a look at KVO and bindings. You can add an observer to 
> any model object (even to object that was compile before KVO was 
> implemented) and when any accessor is call on this model object, 
> the observer will receive a notification.


Likewise .NET.  Most bindable objects implement the appropriate 
"PropertyChanged" event for notification, but of course an object 
that only has the property can go through a notifier proxy that 
handles the same work.

So, in this case, the .NET fallback is more like Cocoa's standard 
approach, while the more typical case has more direct support 
straight from the object itself.

Again, there are subtle differences, but the two environments offer 
basically the same behavior with very little difference to the end-
user writing the code.

Pete