async — Asynchronous vs synchronous client applications This client library supports two modes of operation. These are referred to as synchronous and asynchronous modes. If your application calls MQTTClient_setCallbacks(), this puts the client into asynchronous mode, otherwise it operates in synchronous mode.
In synchronous mode, the client application runs on a single thread. Messages are published using the MQTTClient_publish() and MQTTClient_publishMessage() functions. To determine that a QoS1 or QoS2 (see Quality of service) message has been successfully delivered, the application must call the MQTTClient_waitForCompletion() function. An example showing synchronous publication is shown in Synchronous publication example. Receiving messages in synchronous mode uses the MQTTClient_receive() function. Client applications must call either MQTTClient_receive() or MQTTClient_yield() relatively frequently in order to allow processing of acknowledgements and the MQTT 'pings' that keep the network connection to the server alive.
In asynchronous mode, the client application runs on several threads. The main program calls functions in the client library to publish and subscribe, just as for the synchronous mode. Processing of handshaking and maintaining the network connection is performed in the background, however. Notifications of status and message reception are provided to the client application using callbacks registered with the library by the call to MQTTClient_setCallbacks() (see MQTTClient_messageArrived(), MQTTClient_connectionLost() and MQTTClient_deliveryComplete()). This API is not thread safe however — it is not possible to call it from multiple threads without synchronization. You can use the MQTTAsync API for that.