Re: nil pointers
Re: nil pointers
- Subject: Re: nil pointers
- From: Bill Bumgarner <email@hidden>
- Date: Tue, 11 Jul 2006 15:52:23 -0700
On Jul 11, 2006, at 3:42 PM, Uli Kusterer wrote:
Bill Bumgarner wrote:
You could likely also go for ((id) -1)-- whatever the magic cast
would be-- as that should produce an address that cannot possibly
be a pointer to anything as it is the very last byte of address
space (no room for the other 3 bytes of the address).
I think you mistyped here. No room for more than the first byte of
the object. The address will always have enough memory because
you've likely already allocated an ivar or local to hold it.
The first slot of an object is always a pointer to the object's
metaclass -- the structure that holds the class information -- and
the above was an offhand suggestion that initializing a pointer to
the very last address in the address space is going to guarantee that
any reference through the isa "pointer" is going to explode most
spectacularly.
I don't remember why that was any better than the original suggestion
of 0x1. :-)
As per using some nonsense number in the middle of the address space...
... one shouldn't really be mixing and matching between integral and
pointer types. It is just begging for trouble. Pointer types
always show up as 0x###### and, thus, are readily distinguishable
from integral types.
In any case, this conversation has gotten a bit pendantically
tangential to Cocoa-Dev. Just toss a consistent non-sense number
into your pointers upon initialization.
b.bum
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden