as the spec says, an MQTT-S client may only have one qos1/2 publish in
flight at any one time. A client implementation needs to honour that
requirement so it has the following choices:
1. the easiest thing for a client to do is to discard any newly
submitted publishes until the flow completes - leave it to the
application to deal with resubmitting if it needs to,
2. the client could block the call to submit the publish until the
outstanding flow completes - but it may not be appropriate to block
the application like this,
3. thie client could queue the publishes up internally and send it
when the flow completes - but it means the application won't know
if/when the message is eventually sent.
There is not a default behaviour; it is up to the implementation to
decide which approach to take. Each of which has advantages and
disadvantages.
Regards,
Nick
> --
> To learn more about MQTT please visit http://mqtt.org
>
> To post to this group, send email to mq...@googlegroups.com
> To unsubscribe from this group, send email to
> mqtt+uns...@googlegroups.com
>
> For more options, visit this group at
> http://groups.google.com/group/mqtt