[rabbitmq-discuss] RabbitMQ 2.7.0 is released

[Matthias Radestock]

Stefan,

On 19/11/11 05:52, Stefan Kaes wrote:
> it appears that the FAQ page describing the new reject semantics has
> disappeared.

Indeed it has.

> Here's my current understanding:
>
> Let's say we have a message queue containing messages<m2, m1>, where m1
> will be delivered next.
>
> Let's further assume we have a prefetch count of 1 on our single consumer,
> which is subscribed to the queue.
>
> Then, if the consumer rejects the message, it will be resend to the consumer
> before m2 is sent, right?
>
> In other words: from a consumer perspective, "reject(requeue=true)" behaves
> exactly like "recover" in the scenario above.
>
> Correct?

The second message the consumer receives could be either m1 or m2. 
Arguably that is surprising behaviour so I will raise a bug to look into 
this.

Regards,

Matthias.