Just for fun I write out the actual duration between compute() calls. This is a new/fast EDGE 1050, not running a turn-by-turn route (which can take some processing time). This was a nice 2 hour ride outside. Here is the entire ride, and a 4 min zoom. Like a heartbeat, the execution loop time varies. No problem at all. The actual AVERAGE delay between calls to compute() 1.0000 seconds. The nominal variance was 4-5% +/-. So, a non-issue. It would be interesting to see this for an older model running a route and some CIQ fields. The interesting pattern may be a heisenbug LOL (the act of measuring is causing the artifact) if the FIT writes are queued up and flushed every so often. But that doesn't explain the 950ms iteration, unless the scheduler tries to make up for the delayed call by with a matching advance... which it kind of looks like.
