Github user gemmellr commented on the issue:
https://github.com/apache/activemq-artemis/pull/2113
> So RH docs shouldnt influence the upstream/apache.
On this specifically, the docs pointed to are for an implementation of the
AMQP 0-x style topic binding syntax the Artemis docs indicate its wildcard
syntax is similar to, it seems highly appropriate information to reference and
consider. There is no upstream-downstream relationship between the two in this
case.
More generally, if the expected Artemis behaviour was understood to be the
same as when using the AMQP 0-x syntax it is documented as 'similar to' then
the bug report seems valid. I expect testing a comparable scenario with either
of the Apache Qpid brokers (of which one is actually upstream from the linked
MRG broker behaviour) implementing AMQP 0-x would give the behaviour being
sought. If it isnt understood to give the same behaviour, then it could simply
be decided that the Artemis docs are loose/inaccurate to its implementation and
theyd just need updating to better explain how # wildcarding actually operates.
I agree that fixing this might give a subtle change in behaviour not
expected by some, and so even considering it as a bug would probably keep it
for a 2.7.0 release personally as a result. Certainly theres no need to rush it
into 2.6.1, other version numbers are available and can be used.
(I haven't looked at the code, just commenting on the behaviour)
---
