[ 
https://issues.apache.org/jira/browse/QPID-7649?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16029330#comment-16029330
 ] 

Rob Godfrey commented on QPID-7649:
-----------------------------------

{quote}
It is not clear to me if "Note that if this flag is set to true then the 
endpoints MUST detach and reattach at least once in order to send new 
deliveries." means that the Broker is required to send a detach after actioning 
the unsettled map.
{quote}
The quote from the spec is simply stating the implications of the definition.  
Since new deliveries can't be sent until all unsettled state has been 
exchanged, and the unsettled state could not be exchanged in a single attach 
frame, then at least one more attach frame is needed, which will require a 
detach/reattach.  As an implementation strategy having the server detach 
automatically might seem sensible, but I don't believe it is required.  Having 
said that I think the only way to implement this without having to deal with 
weird stuff like a simultaneous attempt at link stealing, would be to have the 
server simply keep sending attach-detach pairs until the point at which all the 
unsettled map has been sent, whereupon the last attach will have 
incomplete-unsettled=false


> [Java Broker] Support AMQP 1.0 Attach with incomplete-unsettled=true
> --------------------------------------------------------------------
>
>                 Key: QPID-7649
>                 URL: https://issues.apache.org/jira/browse/QPID-7649
>             Project: Qpid
>          Issue Type: Bug
>          Components: Java Broker
>            Reporter: Lorenz Quack
>             Fix For: qpid-java-broker-7.0.0
>
>
> The AMQP 1.0 spec (2.7.3) allows to send an Attach with an incomplete 
> {{unsettled}} map together with {{incomplete-unsettled=true}}. This is useful 
> in cases where the unsettled map is too large to fit in a single frame.
> We currently do not respect this, violating the spec.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to