It is proton that sends the disposition, as you assumed. The spec allows for dispositions to be sent after links have detached, as the deliveries can still be 'live' in some of these cases, but when it occurs after the link is closed like below they actually have no effect sincee the deliveries arent 'live' anymore as the link/terminus is already gone. I wouldnt try to mimmick the behaviour, just delete that test.
On Sun, 22 Sep 2019 at 17:14, Krzysztof <[email protected]> wrote: > > Hi guys, > > We have an issue in NMS-AMQP > https://issues.apache.org/jira/browse/AMQNET-605 which simply results in > prefetched messages not being released after consumer closes down. I looked > at it today, and it seems to be not as simple to fix as I expected. > We have one failing unit tests which was written based on qpid-jms > implementation --> > https://github.com/apache/activemq-nms-amqp/blob/65989e154c34c67b284c91b861d0ddc0a7c47b69/test/Apache-NMS-AMQP-Test/Integration/ConsumerIntegrationTest.cs#L235-L294 > > Can you please help mi to understand how it is possible to send release > disposition after receiver link is closed? From what I've seen in qpid-jms > there is no explicit disposition being send while receiver link is being > closed. Therefore I assume proton must be sending this disposition > implicitly. If so, my question is, whether this behavior meets the amqp > spec. If so, I assume that AmqpNetLite which is our equivalent of proton, > should mimic this behavior. > > Thanks, > Chris
