Let's say I have a class declared as:
@class SomeClass
@interface SomeClass: NSObject {
NSString *myString;
NSString *yourString;
}
@end
And later, in some other code I say:
SomeClass *myClass = [[SomeClass alloc] init];
How does SomeClass know how much memory to allocate given that it didn't override +alloc? Presumably it needs storage for the ivars myString and yourString, but it's using +alloc inherited from NSObject. Is there reference material that covers these details?
+[NSObject alloc] resolves to a call to the Objective-C runtime, which knows the size of each class.
The alloc method does the equivalent of sizeof when it goes to create a new object. Also, the myString and yourString instance variables don't really have too much storage: they are merely pointers, so alloc just allocates enough memory to hold the addresses.
+alloc is just a class method like any other. The default implementation in NSObject, uses class_getInstanceSize() to get the instance size that should be allocated. The instance size is determined based upon a combination of per-class (without inheritance) compile time structure size and per-runtime calculation of total size of the class and all superclasses. This is how non-fragile iVars are possible in 64 bit and the iPhone runtimes.
Some classes, class clusters in particular, don't actually do the true instance allocation until the initializer is called (which is why it is important to do self = [super init] in your initializer methods).