Re: isKindOf: - GSObjCIsKindOf

[Stefan Böhringer]

On 25 Aug 2001 11:06:49 -0400, Marko Mikulicic wrote:
> Pascal Bourguignon wrote:
> > Marko Mikulicic <mikulici@die.supsi.ch> wrote:
...
> >>I have noticed that those methods in NSObject are defined so:
> >>
> >>
> >>- (BOOL) isKindOf: (Class)aClassObject
> >>{
> >>   return [self isKindOfClass: aClassObject];
> >>}
> >>
> >>+ (BOOL) isKindOfClass: (Class)aClass
> >>{
> >>   if (aClass == [NSObject class])
> >>     return YES;
> >>   return NO;
> >>}
> >>- (BOOL) isKindOfClass: (Class)aClass
> >>{
> >>   Class class = GSObjCClass(self);
> >>
> >>   return GSObjCIsKindOf(class, aClass);
> >>}
> >>
> >>
...
> > 
> >          [MyFirstClass isKindOfClass:MySecondClass]
> 
> That doesn't work!
> 
> assert(GSObjCIsKindOf(NSClassFromString(@"ICalendarDate")
>                          ,
>                            NSClassFromString(@"NSCalendarDate")));
> 
> assert([NSClassFromString(@"ICalendarDate")
>                          isKindOfClass:
>                            NSClassFromString(@"NSCalendarDate")]);
> 
> fails the second assertions.
The behaviour of the second assertion is correct with respect to the
OPENSTEP spec. +/-isKindOfClass: test wether the classes are identical.
Therefore [subClass isKindOfClass:classOfWhichSubClassIsASubClass]
returns NO. To test for the subClass relationship no OPENSTEP method
exists. You may search the OMNI-Archives (omnigroup.com) to find a
replacement implementation. 
However, I wonder why -isKindOfClass: is not implemented as follows: 
- (BOOL)isKindOfClass:(Class)aClass 
{ 
   Class class = [self class]; 
   return [class isKindOfClass:aClass]; 
}

Best wishes,

Stefan