[
https://issues.apache.org/activemq/browse/AMQNET-245?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=59474#action_59474
]
Andreas Ländle commented on AMQNET-245:
---------------------------------------
Please Note: Since NMS has dependencies between the NMSTimestamp and
NMSTimeToLive properties - Disabling the timestamps on the producer results in
corrupted expiration dates of the messages!
Reproduction:
Set the TTL of a message to 1 hour - send this message via a producer which has
DisableMessageTimestamp=true.
Now take a look at the "expires" field of the transfered message - it's value
is all to small (and so the broker would discard the message immediately).
Workaround:
I have worked around this issue with a dirty hack - i just bypass the
TTL-Timespan.
var stompMessage = nmsMessage as BaseMessage;
if (stompMessage != null)
{
stompMessage.Expiration = DateUtils.ToJavaTimeUtc(myExpirationDate);
}
> Stomp.MessageProducer doesn't set the TimeStamp to zero (unix epoc) if
> DisableMessageTimestamp is true.
> -------------------------------------------------------------------------------------------------------
>
> Key: AMQNET-245
> URL: https://issues.apache.org/activemq/browse/AMQNET-245
> Project: ActiveMQ .Net
> Issue Type: Improvement
> Components: Stomp
> Environment: Win7, VS2008, .netcf-2.0
> Reporter: Andreas Ländle
> Assignee: Timothy Bish
> Priority: Trivial
> Fix For: 1.3.0
>
> Attachments: MessageProducer.patch
>
>
> If I interpret the comment on Apache.NMS.IMessage.NMSTimestamp correctly, the
> timestamp of a message of send by a message producer with disabled
> timstamping should reflect the unix epoc. This is also what the JMS
> specification says in section 3.4.4 (JMSTimestamp).
> But since the TimeStamp property of a message is set to DateTime.UtcNow in
> Apache.NMS.Stomp.Commands.Message.ctor() and the
> Apache.NMS.Stomp.MessageProducer.Send(...) implementation only updates the
> message timestamp (if message timestampng isn't disabled for the producer),
> the timestamp of a message send via STOMP can never be "zero" (=unix epoc).
> Maybe a simple solution might be to add an else-branch to the
> MessageProducer-Send-implementation (see attached patch file), but i don't
> know if this follows the NMS-projects design intentions.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
