However, it seems not suitable for that case that the messages are ordered by timestamps.
Here is my scenario that needs this kind of feature.
Suppose you have a real time image processing service, starting from many camera stream producers (P), then processed by processors (W) and finally sink to consumers (C).
Processors (W) are stateless. It can benefit from data parallelism by launching multiple W sharing one worker queue.
Any one consumer (C) fetches a specific camera stream passing thru W, requiring that the input stream should be ordered by timestamp.
However, if the processor W latency varies, it does not guarantee the output stream is timestamp-ordered,
because it's possible that partial of specific stream is processed by distinct processors W.
It let me wonder if RabbitMQ can establish priority queues among W and C, that sort the messages by timestamps.
The x-max-priority option would not be relevant here since it has 255 limit.