absurdfarce opened a new pull request, #583: URL: https://github.com/apache/cassandra-cpp-driver/pull/583
Multiple steps to try to keep us out of the crash situation. First: remove the call to the callback from the handler's on_close() function. It's not especially useful when a connection is in the process of closing down like this; when that happens host up or host add events (the objective of the callback in question) should be considered unreliable anyway. This change alone is _probably_ enough to fix the problem but it's effect there is entirely one of timing; by not having to handle the extra event delivery logic we're not taking as long in the callback so there's less opportunity to overlap with other resources being destructed. That said, minimizing the amount of work here is still a pretty good idea. Second: don't make the on_close() function of the handler a pure virtual function. There's no obvious reason this function has to be a _pure_ virtual function so the hope is that by giving it a (no-op) body we'll just execute that if we wind up with a conversion from a PrepareHostHandler to a ConnectionListener object while the event loop is shutting down. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
