Torsten Radtke wrote:
I observed excessive memory usage (working in a garbage-collected environment under 10.6) if I add an observer to a KVO-compliant object. So if I add an observer to say 100000 objects, Instruments shows that 100000 2KB blocks of memory are allocated (additionaly there are another 100000 48B blocks and 200000 32B blocks).
Is this when the first observer is added to each of the 100k objects?
Additionaly, if I remove the observer only 100000 32B blocks are released and collected. This also does not seem right to me.
Suppose that each observed object is doing lazy instantiation of the internal objects needed for managing its observers. What would you expect to see?
The first thing I'd expect is that the cost of adding the first observer would be much higher than adding a 2nd or 3rd observer. The second thing I'd expect is that removing all observers would leave the internal objects in place, because the observed class can't know you won't add another observer in the future.
In other words, the baseline cost of managing observers is paid once, when the 1st observer is added. After that, only the marginal cost of each observer is added or removed as each observer is added or removed.
If you are unhappy about the memory costs of managing observers, then you should probably investigate alternatives. Depending on the class of the observed object, you may be able to save memory, but you have to weigh that against the cost of developing and maintaining your own code, and whether it has or needs the same generality as KVO- compliance. That may or may not be a good engineering tradeoff. YMMV.
-- GG _______________________________________________ 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: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
