[
https://issues.apache.org/jira/browse/DISPATCH-786?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16049664#comment-16049664
]
Alan Conway commented on DISPATCH-786:
--------------------------------------
This is a bug in dispatch: AMQP allows empty annotations to be encoded in two
ways, by omitting the annotation property OR by sending an annotation property
with an emtpy map value. The commit identified above caused the C++ binding to
use the empty-map encoding instead of missing-property. Dispatch should be able
to understand both and treat them as equivalent.
PROTON-1502 has restored the C++ binding to always use missing-property rather
than empty-map. However we still need to fix dispatch to work with the previous
proton C++ client since it should handle both.
This proton commit is known to provoke the dispatch bug:
95491a0 NO-JIRA: c++ enable SSL example tests.
> Message annotations transmitted on wire incorrectly
> ---------------------------------------------------
>
> Key: DISPATCH-786
> URL: https://issues.apache.org/jira/browse/DISPATCH-786
> Project: Qpid Dispatch
> Issue Type: Bug
> Reporter: Kim van der Riet
> Assignee: Chuck Rolke
> Attachments: dispatch-jms.pcapng, qdrouterd.conf
>
>
> Using today's master of dispatch router, qpid-interop-test's JMS messages
> test hangs. A look at a capture of the wire during this test shows the
> following:
> 1. Once the message annotations section is sent, a fragment of the annotation
> just sent is repeated, which is not correctly interpreted on the wire by the
> receiver.
> 2. Dispatch then closes the connection with a "amqp:connection:framing-error"
> the client follows with a TCP RST. However, dispatch appears to then continue
> and sends a flow, disposition and heartbeats to the closed connection.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
