Graphing

The topic is the chart

A graphable publish has five segments. The first two route the message; the last three describe the chart:

username / deviceID / type / graphName / seriesName
   larry / a8f3k2m1 / numb / temperature / celsius   ->  72.5

The type tells the platform how to plot it. The graphName is the chart's title. The seriesName is one line on that chart. Send a value and the chart, the axis, and the line all come into existence on their own — no configuration, no schema, no registration step.

Number graphs and boolean graphs

The type segment picks the kind of graph you get:

For the full list of payload types and what each one does, see Data Types & Routing.

Multiple lines on one graph

Reuse a graph name with different series names and the lines land on the same chart:

// two lines on the "temperature" graph
mqtt.publish("~/~/numb/temperature/indoor",  72.5);
mqtt.publish("~/~/numb/temperature/outdoor", 58.3);

Each distinct series name is its own line, colored automatically. There is no limit you need to declare ahead of time — add a new series name and a new line appears.

History and retention tiers

Number data is rolled up into time buckets as it ages, so a chart can show the last few minutes or the last few years without dragging every raw point around. The platform keeps six tiers — minutes, hours, days, weeks, months, years — and holds up to the 3,000 most recent points in each. When you zoom the dashboard chart out, it reads from a coarser tier; when you zoom in, a finer one.

Sending more in one message

The five-segment form sends one value per publish, which is the clearest place to start. When you need to send several readings at once — a sleepy battery device, a high-frequency publisher — you can move structure out of the topic and into a JSON payload. The graphs come out identical. See Topic Flexibility for the five shapes the same data point can take.