On Oct 17, 2011, at 10:19 AM, Gerriet M. Denkmann wrote:
> I understand this. So I tried to tell the compiler that the passed selector
> is not like copy:
>
> - (void)replaceIn: thing readSelector:
> (__attribute__((ns_returns_not_retained)) SEL)selectorIn
> {
> NSString *key = [ thing performSelector: selectorIn ];
> }
>
> But it still complains: "PerformSelector may cause a leak because its
> selector is unknown".
> Sure, it is unknown, but it's non-retaining behaviour is known.
The attribute you added is not applicable here, so it is ignored (the attribute
must be added to a function or method declaration, where as a SEL is just a
data type). Consider also that you can't declare such a thing universally for
all SEL – it is the underlying method on a particular class that has the
behavior you are looking for.
> Is this a bug?
No, its the compiler admitting that it has no way to verify that what is
happening will have the behavior you expect. Thats why this is a warning and
not an error.
--
David Duncan
_______________________________________________
Cocoa-dev mailing list ([email protected])
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:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to [email protected]
