I looked at work @bptworld did, and it was too complicated for me and when looking at latency I don't like complicated, and I like visuals better. Like this:
So I put a very simple driver together than you can use for really simple timing of anything. You send it an On, and then later send it an off and it records the time between the two events. It reports a "pressure" event in kPA of the number of milliseconds between the two. Why pressure? I was fixing up the InfluxDB app*, and it naturally exports pressure sensors well, and I figured it was something I could overload without messing up to much other data, and it's unlikely that most people have lots of pressure sensors around. So shrug I'm open for ideas for a more appropriate sensor value type.
There's also an auto more where it sets a default 50ms runIn, where you can see how accurately the hub manages to run the off in 50ms.
For example to time an RM4 rule you could do: (ok it's a stupid example, but you get the idea):
Or this slightly complicated option, which surely could be better:
**) I've updated the InfluxDB-Logger thread with a link to a version that does write combining to avoid 20-50 async https calls/sec, and keeps stats to see if there are issues.