With regard to the last point in the talk where Walter was suggesting not calling finalizers on objects whose code has been unloaded - would it not make more sense to simply call all the finalizers before unloading the library?

If the finalizer is not called you will potentially get resource leaks - although there's no guarantee currently that it will be called, isn't there the assumption that if it's not called the object still exists, even if it's not referenced anywhere? ie. you still have some guarantee that if you keep allocating objects and then unreferencing them that the number of unfinalized objects will never exceed some large but fixed value.

Reply via email to