MQTT dialects
Like HTTP, MQTT is "just" a protocol, which defines how events or messages are being transported over the wire. On top of that, oen needs to define an API. Some might disagree and say that you can just publish anything to an MQTT broker. However, implicitly, you have to agree (even when only working with your own components) to a data format and topic structure.
Introduction
As Drogue Cloud is payload agnostic, we don’t interfere or interpret the payload itself[1].
However, we sometimes need it interpret the metadata of events and messages, like the topic a client publishes or subscribes to.
The MQTT endpoint supports multiple dialects. However, it also needs to know which dialect a device "speaks". Using MQTT v5, it would be easy to add some property during the connection phase, to indicate which dialect the device expects. However, as MQTT v3.1.1 still seems to be the dominant protocol version, that isn’t possible[2].
That is why Drogue Cloud keeps this configuration as part of the application or device configuration. The advantage of that is also, that it can be centrally managed. Just assume some protocol setting of your device, which is somewhere out in the field, is wrong. Being able to correct that in the cloud, might be beneficial.
Basic configuration
The configuration for MQTT dialects is located in either the application or device resource at .spec.mqtt.dialect.
The configuration of a device will override the configuration of an application. If neither the application nor
the device has any configuration, the drogue/v1 dialect will be used.
spec:
mqtt:
dialect:
type: plainTopic (1)
devicePrefix: true (2)| 1 | Selects the dialect. |
| 2 | Additional options on the same level. |
Dialects
The follow section describes the configuration of the different dialects. To get an understanding of the different dialects themselves, see Protocol dialects.