Enqueued qos0 dispatch MQTT_EVENT_PUBLISHED (IDFGH-12511)

[nebkat]

Dispatch MQTT_EVENT_PUBLISHED when an enqueued QOS0 message is actually published.

[euripedesrocha]

Hi @nebkat thanks for your contribution. Could you clarify your use case here?

[nebkat]

Hi @euripedesrocha!

When you esp_mqtt_client_enqueue a qos0 message it gets added to the outbox but there is no confirmation that it actually got sent to the broker.

Qos1+ can be used to receive the explicit confirmation back but I do not need that level of guarantee, I just want to know when the message got sent (including for tracking # of messages successfully published in this session).

esp_mqtt_client_publish also doesn't work for me because I'm publishing in an event handler than needs to return quickly.

And lastly, I would like to experiment with semaphore synchronized publishing which would be easier if all messages triggered MQTT_EVENT_PUBLISHED.

[euripedesrocha]

@nebkat thanks for clarification. We don't have the event for QoS 0 messages because it at first don't make sense given that they are a QoS level that is used when user don't want to track the message. Could you clarify why you want to have the synchronized publishing?

[nebkat]

These are technically two separate issues - with QoS 0 we might not care whether the broker actually successfully processes the message once it is received, but we may still want to know whether the client ever attempted to send the message in the first place (since we queued it for asynchronous sending and otherwise have no way of knowing).

In other words: it is ok if my broker or network connection somehow lose the message, but it seems wasteful to have to depend on the broker to observe the behavior of the client (by upgrading to QoS 1/2).

With this change I use MQTT_EVENT_PUBLISHED to count the total number of messages and bytes published across all QoS.

For synchronization, I would like to be able to do:

// in handler
case ESP_MQTT_PUBLISH:
    xTaskNotifyGive(task_handle);

// in task
esp_mqtt_client_enqueue(...); // QoS0
// Possibly do something here
if (!ulTaskNotifyTake(true, pdMS_TO_TICKS(1000))) return; // Stop publishing batch if previous message fails
esp_mqtt_client_enqueue(...); // QoS0
// Possibly do something here
if (!ulTaskNotifyTake(true, pdMS_TO_TICKS(1000))) return; // Stop publishing batch if previous message fails

This is not possible with esp_mqtt_client_publish. Furthermore, I believe using esp_mqtt_client_publish will skip ahead of any enqueued messages which is undesirable for me.