Updates:
Status: WorkingAsIntended
Comment #7 on issue 3523 by [email protected]: Deadlock after calling
IdleNotification(100) or Dispose on isolate
https://code.google.com/p/v8/issues/detail?id=3523
I don't want to wait with the deallocation of my C++ instances
You don't have to. That's what weak callbacks are for. Deallocating C++
objects when you get a weak callback for them is fine. *Relying* on getting
that callback before shutdown is wrong.
hope that when it would be necessary V8 would call the garbage collector
and until then just ignore [the valgrind reports]
It most certainly will, but the garbage collector can't clean up your C++
objects. The only reason it looks like the GC affects Valgrind's output is
because the weak callbacks are creating a connection between otherwise
separate systems.
As far as I understand the situation, you should keep your weak callbacks
so you can do early cleanup when appropriate, and then you'll need an
additional cleanup mechanism, entirely on the C++ side, that cleans up
everything before program shutdown, regardless of whether weak callbacks
have been fired or not. Roughly:
global list_of_things;
CreateThing() {
register thing in list_of_things;
install weak callback on thing;
}
WeakCallback(thing) {
remove thing from list_of_things;
clean up resources held by thing;
}
Shutdown() {
for (thing in list_of_things) {
clean up resources held by thing;
}
}
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
