On May 14, 2013, at 12:53 AM, Roland King <r...@rols.org> wrote:
[...]
>> - (void) mouseDown: (NSEvent*) event
>> {
>> NSView* superView = [self superview];
>> [self removeFromSuperview];
>> // [superView addSubview: self];
>> }
>>
>> ...then dealloc does in fact get called. But if I uncomment that one line,
>> which references self, dealloc does not get called. This is good news -- I
>> would want ARC to work regardless of whether it calls non-ARC code, and vice
>> versa.
>>
>
> well you have several lines in there which reference self, [ self superview ]
> and [ self removeFromSuperview ].
I was thinking that if calling [self removeFromSuperview] removed the last
reference to self, dealloc would have been called in both cases. Naturally I
wouldn't expect self to have been dealloc'ed before that statement.
> The only difference in the last one is that self is used as a parameter to a
> call to another object and that may possibly be the difference.
I fiddled a bit with it, adding calls that don't take arguments like [self
description] and [self self]. There was no dealloc until just before the method
exited, after those calls. It *seems* that merely referring to self, whether as
receiver or argument, keeps it alive. However, I might just have been lucky.
> And the LLVM documents talk about this, they say that self is not retained
> during a method call.
>
> http://clang.llvm.org/docs/AutomaticReferenceCounting.html#self
>
> and discuss exactly this risk.
Thanks. Just goes to show it's dangerous to guess too much -- in my case, that
self is a strong reference -- from too little evidence.
--Andy
_______________________________________________
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
https://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
