Re: Why is this casting needed?

[Helge Hess]

On 25. Mrz 2006, at 12:51 Uhr, Günther Noack wrote:
> Am 25.03.2006 um 00:57 schrieb Adrian Robert:
> > But it's only a warning -- it still works at runtime in either  
> > case.  What is the compiler doing differently with the cast?  Is  
> > it some performance optimization?
> The resulting code doesn't differ.

If you have those methods:

  @implementation C:
  - (void)addNumber:(float)value;

  @implementation A:
  - (void)addNumber:(short)value;

  @implementation B:
  - (void)addNumber:(NSNumber *)value;

The generated code of course _does_ differ. How exactly it differs  
depends on the C ABI of the host system.

This is why its recommended to explicitly include the type in the  
method for base type stuff, like:
  - (unsigned int)unsignedIntValue;
  - (id)initWithUnsignedInt:(unsigned int)value;
etc.


The problem is a bit less when both methods take objects as  
arguments, like:

  @implementation C:
  - (void)addNumber:(NSNumber *)value;

  @implementation A:
  - (void)addNumber:(NSValue *)value;

Of course in this case the generated code is the same. However, the  
static type checking being applied is the same in the C compiler. And  
the reason why the compiler needs to do type checking is also the same.

If you want to accept arbitary objects with your selector, use just
  - (void)addNumber:(id)value;

Which is the 'traditional' ObjC way :-)

Greets,
  Helge
--
http://docs.opengroupware.org/Members/helge/
OpenGroupware.org