Just a small tweak to my suggested pattern for having -init and -initWithCoder
share implementation.
Maybe the shared code should be in a method, but if it is, we have to ensure
that the methods are never overridden and each supercalss/subclass has a unique
one. To ensure this we could require the method be given a name like
-(id)_CurrentClassNameInit.
Also I realized the method should return self, so if there is some fatal error,
it can release itself and cause the real -init or -initWithCoder to return nil.
So, here's my updated pattern suggestion:
@implementation MyClass
- (id) _MyClassInit
{
self->someCArray = malloc(100);
self->someValue = 123;
...
if (someFailure)
{
[self release];
return nil;
}
return self;
}
- (id) init
{
self = [super init];
self = [self _MyClassInit];
return self;
}
- (id)initWithCoder: (NSCoder*)coder
{
self = [super initWithCoder: coder];
self = [self _MyClassInit];
if ([coder containsValueForKey: @"foo"]) [self setFoo: [coder
decodeObjectForKey: @"foo"]];
if ([coder containsValueForKey: @"bar"]) [self setBar: [coder
decodeObjectForKey: @"bar"]];
...
return self;
}
@end
--Eric
_______________________________________________
Gnustep-dev mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/gnustep-dev
