On Sep 23, 2004, at 10:28 PM, Dave Deakins wrote:
"Doug Gregor" <dgregor@cs.indiana.edu> wrote in message news:D5FEB638-0DB1-11D9-A949-000D932B7224@cs.indiana.edu...
On Sep 5, 2004, at 5:51 PM, Dave Deakins wrote:
...should there be a:
watch_bound_objects.set_controlling();
statement at the very end of the slot_base::create_connection() procedure (after the safe_connection.release() statement) in slot.cpp?
I don't think that's the right solution, because watch_bound_objects gets copied around a bit, and would then disconnect too early. I'm revisiting this code now to try to isolate the bug.
I agree. After trying out that idea, I noticed exactly what you said. It did indeed cause slots to disconnect prematurely. However, what if you made watch_bound_objects a shared_ptr<connection> instead of a just a connection? In that case, you could still set it as controlling at the end of create_connection, but the connection wouldn't be destroyed unless all copies of the slot ceased to exist. It seems like that would work to clean up the slot/trackable references at the proper time, but maybe there is a more elegant solution than that.
-Dave
I'm sure there's a more elegant solution, but I'm a bit afraid to rock the boat with Aleksey trying to roll a release :) I'm going to give myself the evening to see if I can think up any problems with this solution, and will try to implement it within the next day or so. Thanks for your help! Let's hope this thing cleans up nicely :) Doug