Re: Re[6]: Recent key-value encoding changes to NSObject.

[Richard Frith-Macdonald]

On Monday, February 11, 2002, at 05:54 PM, Manuel Guesdon wrote:
> > | I suggest that you do as follows in your custom subclass -
> > |
> [...]
> > |   /* If there is an ivar, use it.  */
> > | This should support all your needs -
> > |
> > |  ivars take precedence over everything else.
> > |  dictionary values take precedence over getXXX.
> > |  getXXX are still used when all else fails.
>
> But getXXX should be called before ivar access (during the valueForKey 
> process). :-(

Sorry if it sounds bad, but I think you are too close to this and need 
some perspective.
Nicola just provided an example ... the point really is that you can 
quite easily do
whatever you want to.  If you want to use the private get methods first, 
that's fine.

so your method would be something like this -

- (id) storedValueForKey: (NSString*) aKey
{
  id result;

  if (private_get_method_exists(aKey) || private_ivar_exists(aKey))
    return [super storedValueForKey: aKey];
  result = [_values objectForKey: aKey];
  if (result == nil)
    return [super storedValueForKey: aKey];
  return result;
}

pretty simple really - and the code duplication with gstep-base would be
the stuff to test to see if a method or ivar exists ... simple string 
manipulation
to make the appropriate names, and calls to respondsToSelector: or 
GSInstanceVariableInfo()

The only downside of doing it that way is a little inefficiency in 
testing some things twice
in some cases - once in your implementation, and again in the superclass 
implementation.