> Does this mean that there is no guaranteed way to ensure that a
> timer will never fire again when its target is going away? (Or is
> this simply a bug that I should write up.)
Nevermind. I should have done more debugging first. I created a
retain cycle between my target and my timer, invalidated the timer
(which was holding the last reference to self) at the beginning of a
method, then used the object aftewards. My fault...