What’s the difference beetween NSCoder’s encodeObject: and encodeObject:forkey:?

Question

In NSCoder, you can call encodeObject: and encodeObject:forkey:. And this for many data types. What is the difference beetween those calls? How to use them? Why isn't there a encodeDataObject:forkey: or encodePropertyList:forKey:?

Answer 1

Keyed vs. Unkeyed Accessors

Most of the time you just call the encodeSomething:forKey: method and supply a key that you later use to get the value back from a decoder:

- (void) encodeWithCoder: (NSCoder*) coder
{
    [coder encodeObject:someProperty forKey:@"someProperty"];
}

- (id) initWithCoder: (NSCoder*) decoder
{
    self = [super init];
    if (self) {
        [self setSomeProperty:[decoder decodeObjectForKey:@"someProperty"]];
    }
    return self;
}

The unkeyed version of the call is older, I guess that serialization used to work differently back before 10.2.

Specialized Object Accessors

There isn’t a separate encodeDataObject: call since NSData conforms to NSCoding, so that you can encode it using regular encodeObject:. The same applies to property lists – a property list is just a dictionary, and a dictionary is a regular object that can be encoded using encodeObject:.

What maybe got you confused is the number of specialized methods for encoding primitive types like BOOL or NSUInteger. These have to do with the type system. When encoding and decoding objects, the interface can use the id type and it works regardless of the particular object type:

// can pass NSObject*, NSData*, any object
- (void) encodeObject: (id) anObject {…}
- (id) decodeObject {…}

There is no such general, “wildcard” type for primitive types, therefore the number of specialized getters and setters:

- (void) encodeBool: (BOOL) flag {…}
- (BOOL) decodeBool {…}

Theoretically you could use void* and cast, but that’s clumsy and the encoding interface wouldn’t know the size of the object to encode anyway.