[
https://issues.apache.org/jira/browse/ARTEMIS-1933?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16523238#comment-16523238
]
Lionel Cons commented on ARTEMIS-1933:
--------------------------------------
WRT the standards...
If there are standards (like JMS) then brokers should comply. If they don't
then this clearly indicates a bug.
If there are no standards then we can argue about which behavior should be
implemented or not. So this blurs the distinction between a bug and a feature.
The bottom line is that Artemis behaves differently from ActiveMQ 5.x (here in
this scenario). So, IMHO, we have three (main) options:
# Artemis is right and ActiveMQ is wrong so the current behavior is a bug in
ActiveMQ
# ActiveMQ is right and Artemis is wrong so the current behavior is a bug in
Artemis
# both Artemis and ActiveMQ are right (i.e. no bugs) and they intentionally
behave differently (which is problematic when people want to migrate from one
broker to the other)
So, which option do you think applies here?
> Wildcard subscriptions deliver queue message multiple times (STOMP)
> -------------------------------------------------------------------
>
> Key: ARTEMIS-1933
> URL: https://issues.apache.org/jira/browse/ARTEMIS-1933
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Reporter: Lionel Cons
> Priority: Major
> Attachments: ARTEMIS-1933.text, broker.xml
>
>
> Here is the scenario (using STOMP):
> - one subscription to {{/queue/test.foo}}
> - one subscription to {{/queue/test.*}}
> - one message sent to {{/queue/test.foo}}
> Since we are dealing with queues, the message should be load-balanced between
> the two matching subscriptions so only one should get it. This is what
> ActiveMQ 5 does.
> With Artemis, both subscriptions get the message.
> FWIW, I'm using {{default-address-routing-type}} to make sure destinations
> starting with {{/queue/}} act like a queue (see ARTEMIS-1906).
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
