I took a brief trip on Monday:
2020-04-20 17:30:00
2020-04-20 19:30:00
(Pacific Daylight Time)
But the detailed trip timeline https://floathub.com/trip/1762
starts at about 15:30 and goes to about 17:30. It's like the details are on CDT. Am I missing something?
The trip process uses the latitude and longitude of the trip coordinates to do a timezone lookup. Which in this case it got right (PDT), but the local time calculation is clearly still 2 hours off (consistent with being two timezones further to the west). Suspect it may be a DST issue (moving an hour the wrong way would be consistent with a 2 hour offset). We will check on this and update you in a bit. Thanks for bringing it to our attention.
Confirmed; in some cases we were subtracting the Daylight Savings Time hour rather than adding it, resulting in a net 2 hour error.
Believe we have resolved this, and changes are now in place.
Thanks again for the feedback, and please let us know if you see any other issues.
Hi FH! The trip detection is getting better (for me) - I see the trip I did yesterday afternoon out into the bay, anchored, and returned. But then I also got a small "trip" inside the marina after that. I'm not sure how the rules are expressed but this one seems easy to rule out.
Thanks for pointing out, we will review. We're still fine tuning the smoothing, but in any case this should have been caught by the "overall distance" filter (the overall distance was trivial).
And...this one's the opposite - it split yesterday's trip into two separate trips even though we were anchored for a little over 3 hours. It was a short trip though, not sure if that impacts the algorithm.
The issue there is that we were tweaking the issue with GPS spidering for accounts with challenging home port GPS readings (metal coverings, etc.). We were tuning the radius up of what it means to be "back home" at the location where the trip started. Where you stopped for your anchor break was within that new, larger testing radius. So when you were stationary (at anchor) and within 500 meters of where you started (large radius), it marked your trip as complete.
Have to tune that upper bound down, but we need to do that as part of better smoothing and outlier detection, which is in progress.
Thanks for looking at this. Normally that wouldn't be an issue but this pandemic thing is changing up boating habits. Meydenbauer Bay was packed yesterday afternoon - it look like 4th of July on Lake Union!
Very valuable to have feedback like this, gives us lots of edge cases to use as we work through the improvements.