Re: [Springframework-rcp-dev] Problem with BeanPropertyAccessStrategy

[Andy D. <an...@ma...>]

On Tuesday 24 August 2004 09:33 pm, Oliver Hutchison wrote:
> The most robust/encapsulation safe solution would be this:
>
>   Object oldValue = beanWrapper.getPropertyValue(aspect)
>   beanWrapper.setPropertyValue(aspect, newValue);
>   if (! ObjectUtils.nullSafeEquals(beanWrapper.getPropertyValue(aspect),
> oldValue) {
> 	fireValueChanged(aspect);
>   }
>

The reason this might break encapsulation is because it assumes all properties 
on all beans must work exactly the same.  The assumption the above code makes 
is: a bean will only ever set a property value to the very same value I 
passed to the setter.  The bean will not perform any additional processing on 
the value and it will not end up setting an alternate value instead of the 
one I passed in.  Granted, this is true in most cases - but I'm sure there 
will also be cases where the value you pass to a setter is not the value that 
ends up in the property (for example, a setter may massage the value or wrap 
it in some other object instance before setting).  Having said all this, I 
still think there is value in the above solution as long as there is a way to 
override it.  That is why I like your idea below...  

> Perhaps as a compromise we could do the above if the bean does not
> implement PropertyChangePublisher and keep the current implementation if
> it does.

I agree that it would be nice not to change my domain objects.  Having the 
option to alter behavior is always nice, and having a default that works in 
most cases is doubly nice.