NOTES

Comments from Sean Treadway, 2 June 2008, on the rabbitmq-discuss list:

 - On naming, extensibility, and headers:

   "STOMP looked like it was MQ agnostic and extensible while keeping
    the core headers well defined (ack=client, message_id, etc...),
    but my application was not MQ agnostic.  Plus I saw some of the
    ActiveMQ headers weren't available or necessary in RabbitMQ.

   "Keeping the AMQP naming is the best way to piggy back on the AMQP
    documentation.  For those that need simple, transient queues, the
    existing STOMP documentation would be sufficient."

   ...

   "I only have experience with RabbitMQ, so I'm fine with exposing
    AMQP rather than try to come to some agreement over the extension
    names of standard STOMP headers."

 - On queue deletion over STOMP:

   "Here, I would stick with the verbs defined in STOMP and extend the
    verbs with headers.  One possibility is to use UNSUBSCRIBE
    messages to change the queue properties before sending the
    'basic.cancel' method.  Another possibility is to change queue
    properties on a SUBSCRIBE message. Neither seem nice to me.  Third
    option is to do nothing, and delete the queues outside of the
    STOMP protocol"

Comments from Darien Kindlund, 11 February 2009, on the rabbitmq-discuss list:

 - On testing of connection establishment:

   "[O]nce I switched each perl process over to re-using their
    existing STOMP connection, things worked much, much better.  As
    such, I'm continuing development.  In your unit testing, you may
    want to include rapid connect/disconnect behavior or otherwise
    explicitly warn developers to avoid this scenario."