iPhone Core Data crash with ARC

Question

This is my first time trying to use both ARC and Core Data. I can't seem to figure out why my code is crashing.

in the .h I have:

@interface Foo : NSObject {
}
@property (nonatomic,strong) NSString *name;
@property (nonatomic,strong) NSString *email;
@property (nonatomic) BOOL myBool;
@property (nonatomic) float myFloat;

in the .m

@implementation User

@dynamic name;
@dynamic email;
@dynamic myBool;
@dynamic myFloat;

User *user = (User *)[NSEntityDescription insertNewObjectForEntityForName:@"User" inManagedObjectContext:[appDelegate managedObjectContext]];

[user setName:[[[dictionary objectForKey:@"user"] objectForKey:@"user"]objectForKey:@"name"]];
[user setEmail:[[[dictionary objectForKey:@"user"] objectForKey:@"user"]objectForKey:@"email"]];
[user setAuthorisation_token:[[[dictionary objectForKey:@"user"] objectForKey:@"user"]objectForKey:@"authentication_token"]];
[user setMyFloat:5]; <------ crash when set to anything other than 0 (eg,setting to FALSE will crash it).
[user setMyBool:FALSE]; <---- crash when set this to anything other than 0.

Basically whenever I try to use a type other than a string I am getting EXEC crash on that particular line. When I use strings for everything it is fine. In my.xcdatamodeld file I have myFloat set to FLOAT and myBool set to BOOLEAN

kill
error while killing target (killing anyway): warning: error on line 2184 of "/SourceCache/gdb/gdb-1708/src/gdb/macosx/macosx-nat-inferior.c" in function "void macosx_kill_inferior_safe()": (os/kern) failure (0x5x)
quit
Program ended with exit code: 0

Answer 1

You can't use @dynamic for primitives (like float and BOOL) because Core Data won't create implementations for them.

So the reason why your code is crashing is because when you use @dynamic you are telling the compiler "I promise that an implementation for these getters and setters will be available at runtime". But since Core Data doesn't create them then your code tries to call methods that doesn't exist.

Instead there are two things you could do: Use an NSNumber for both the BOOL and the float or implement your own getters and setters.

Using NSNumber:
Core Data only uses objects and not primitives but you can specify boolean or float in the Model. When you call [user myFloat] you will actually get an NSNumber back with the float value inside it. To access the primitive you then call float f = [[user myFloat] floatValue];. The same thing goes for the boolean, it also gets stored in an NSNumber. So when you try to access it you will get back an NSNumber that you need to call BOOL b = [[user isMyBool] boolValue]; to get the primitive back.

The same thing goes the other way around, when setting myFloat and myBool, you need to store them inside an NSNumber, e.g. [user setMyFloat:[NSNumber numberWithFloat:f]]; and [user setMyBool:[NSNumber numberWithBool:b]];.

To use this approach you would have to change your last two properties to

@property (nonatomic, strong) NSNumber *myBool;
@property (nonatomic, strong) NSNubmer *myFloat;

but you can keep the @dynamic for both of them.

Implementing you own getters and setters:
For your convenience, you may want your "user" object to get and set the primitive types, float and BOOL, directly. In that case you should keep the properties as float and bool and in your implementation file (.m) remove the @dynamic lines for myFloat and myBool.

To implement the getter and setter you need to know a little about KVC/KVO and Core Data. In short: you need to tell the system when you are about to access or change a property and when yo u are done accessing or changing it, since Core Data won't do it for you. Between the "will access/change" and "did access/change" you are free to retrieve or modify the properties. One more caveat is that Core Data still cannot save the BOOL and float directly, so they need to be packaged into and unpackaged from NSNumbers when getting and setting.

Further, you can't call [self setValue:ForKey:]; or [self valueForKey:@""]; because that would cause the method you are in to call itself and throw you into an infinite loop. Core Data solves this use-case by allowing you to get and set the value without hitting your own implementation by calling [self setPrimitiveValue:ForKey:] and [self primiveValueForKey:]. Note: primiteValueForKey has nothing to do with primitive types (int, float, BOOL) but is just the name of the methods you use to get and set values in Core Data directly.

The implementation for your float and BOOL would look something like this:

- (float)myFloat 
{
    [self willAccessValueForKey:@"myFloat"];
    float f = [[self primitiveValueForKey:@"myFloat"] floatValue];
    [self didAccessValueForKey:@"myFloat"];
    return f;
}

- (void)setMyFloat:(float)f 
{
    [self willChangeValueForKey:@"myFloat"];
    [[self setPrimitiveValue:[NSNumber numberWithFloat:f] forKey:@"myFloat"];
    [self didChangeValueForKey:@"myFloat"];
}

- (BOOL)isMyBool 
{
    [self willAccessValueForKey:@"myBool"];
    BOOL b = [[self primitiveValueForKey:@"myBool"] boolValue];
    [self didAccessValueForKey:@"myBool"];
    return b;
}

- (void)setMyBool:(BOOL)b 
{
    [self willChangeValueForKey:@"myBool"];
    [[self setPrimitiveValue:[NSNumber numberWithBool:b] forKey:@"myBool"];
    [self didChangeValueForKey:@"myBool"];
}