Changes between Version 1 and Version 4 of Ticket #147

Timestamp:

10/08/2011 04:20:30 PM (13 years ago)

Author:

Brian Warner

Comment:

well, in this case, the code exists to handle references being GCed *before* the process shuts down. It's just Python's eagerness to cleanup everything at shutdown (which is counter to your recommended crash-only approach) which happens to be firing our Referenceable's destructors. If we had a way to know that the whole process was being torn down soon anyways, we could change the code to just skip these destructors (no need to send a "you can drop reference 12 now" message when the TCP connection is about to be closed). But I don't know how to detect that.