Objective-C:
NSLog(@"%@", [[NSArray arrayWithObjects:@"a",@"b",nil] objectAtIndex:-1]);
// => NSRangeException: *** -[NSCFArray objectAtIndex:]: index (-1 (or possibly larger)) beyond bounds (2)
Bit-pattern-wise, (int32_t)-1 == (uint32_t)0xFFFFFFFF, so that innocent-but-invalid -1 gets interpreted as ULONG_MAX.
ULONG_MAX is waaaay outside the bounds of our 2-element array, so an invalid index exception gets correctly thrown.
But things can get different in Cocoa Uncanny Valley:
JSCocoa/JSTalk:
[[NSArray arrayWithObjects:@"a", @"b", nil] objectAtIndex:-1];
// => @"a"
Somewhere in JSCocoa’s machinery -1 morphs into zero, and the first element is retrieved sans exception. Ouchie.
Objective-J:
[[CPArray arrayWithObjects:@"a",@"b",nil] objectAtIndex:-1];
// => undefined
Cappuccino’s CPArray doesn’t throw an exception, just quietly returns nothing. I wish it barked like NSArray.
MacRuby:
NSArray.arrayWithObjects('a','b', nil).objectAtIndex(-1)
# => probably NSRangeException
My version of MacRuby (0.5b1 I think) crashes inside rb_vm_print_current_exception(), so I think it’s trying to sputter out NSRangeException.
Nu:
((NSArray arrayWithList:`("a" "b")) objectAtIndex:-1)
;; => NSRangeException: *** -[NSCFArray objectAtIndex:]: index (-1 (or possibly larger)) beyond bounds (2)
Not surprised, Nu is in perfect mesh.
