[
https://issues.apache.org/jira/browse/ARTEMIS-3180?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17304225#comment-17304225
]
Gary Tully commented on ARTEMIS-3180:
-------------------------------------
[~erwindon] fair point. I must admit to being surprised by the 0 or more
semantic for #. The first two do seem implicit, in particular the # on its own
is often used as a default (catch all) address setting.
I am not sure on the last option there, combining with text. That would be
really expensive to compute. I think the matching must be combined to parts or
paths or components, what ever is delimited.
I think one issue we have is that there are numerous places where wildcards are
used and a few places in code where they are parsed. That is not ideal.
For Binding queues to addresses, where either can be a wildcard, the matching
is delimiter based. That was my initial focus.
>From a broker perspective, the simpler the wildcards are the better,
>especially at scale.
There is probably some room for better documentation, however we need to verify
that the documentation is valid for all of the uses of wildcards in the broker
before making any change.
> Consumers with wildcards addresses broken
> -----------------------------------------
>
> Key: ARTEMIS-3180
> URL: https://issues.apache.org/jira/browse/ARTEMIS-3180
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Components: AMQP, Broker
> Affects Versions: 2.17.0
> Environment: Artemis 2.17.0
> openjdk version "1.8.0_275"
> OpenJDK Runtime Environment (build 1.8.0_275-b01)
> OpenJDK 64-Bit Server VM (build 25.275-b01, mixed mode)
> Reporter: Sebastian T
> Assignee: Gary Tully
> Priority: Blocker
> Fix For: 2.18.0
>
> Attachments: artemis-3180.zip
>
> Time Spent: 3h 10m
> Remaining Estimate: 0h
>
> While upgrading from Artemis 2.16.0 to 2.17.0 several of our test cases
> started failing.
> We are using AMQP and QPid JMS to communicate to Artemis.
> The issue seems to be related to subscriptions with wildcards.
> Sending a message to an address "topics.aaa.bbb.ccc" with a consumer
> subscribed to "topics.#.aaa.#" and one with "topics.#.bbb.#" only the
> consumer subscribed to "topics.#.aaa.#" receives the message.
> I am attaching a test case illustrating the issue.
> Running "mvn test" executes it with Artemis 2.17.0 and will fail. "mvn test
> -Dartemis.version=2.16.0" executes the test with Artemis 2.16.0 and will
> succeed.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
