BlockingConnection basic_publish delays exchange not found error until next channel operation

[leth]

If you try to publish a message to an exchange which does not exist, the blocking connection channel will not raise an exception until the next channel operation is attempted.

A workaround seems to be to call process_data_events() on the connection after the channel publish call.

[vitaly-krugl]

@leth, that's the nature of BlockingConnection. If you're not executing code in BlockingConnection, then it's not processing input either. If you're using RabbitMQ that supports Publisher Acknowledgements, then you can enable it on a given channel via channel.confirm_delivery(), and you should then get the "exchange not found" error before basic_publish returns.

[vitaly-krugl]

Works as intended. Let's close.

[vitaly-krugl]

Please see my earlier comments. This feature works as intended in BlockingConnection. SelectConnection and other asynchronous adapters can be more responsive if used appropriately in an I/O-bound applications.