DST and sleep calculation

[Hinfty]

Amazed that sleep time calculation does not seem to take daylight saving time (or the switch back) into account correctly.

Sleeping from 23h to 7:30 the next day is shown as 7h 30min sleep with 1h being awake. So that is missing an hour.

Then I edited that sleep, because one of the detail pages showed that same sleep until 6:30. Edited that to 7:30, which completely broke sleep phases being shown. 

Now deleted and re-added the sleep with same times and now I have 9h 30 min asleep, but magically also 200min awake. The hour with dst change has a completely blank sleep phase.

So it's obvious that dst is handled very poorly by fitbit. 

It seems fitbit logs sleep in dst/non-dst times instead of handling/logging all data in utc and showing the user the times in whatever zone/time currently applies. 

[LZeeW]

@Hinfty   Every time zone change can lead to data loss.  It doesn't matter if it is the start/stop of BST/DST or travel.  Sometimes users must manually reset their timezone, because the automatic timezone setting fails.  A manual reset there shows a dialog box warning of a possible data loss.

[Hinfty]

"every time zone change can lead to data loss" 

Sorry, that's just not true.

A competent IT team should easily be able to work with time zones and time zone changes. Its such a classic example in computer science for how to handle cases.

Example for how fitbit could easily avoid all issues:

-all durations are calculated in UTC timezone/format. No DST, so this always works and is a steady flow of time. 

-the axis in the graphs are shown/drawn in local timezone. For DST change days, this has the time between eg 2 and 3 AM twice or not at all

I hope you can see how DST (or users switching time zones) does not need to lead to data loss, if the backend is programmed reasonably. 

[LZeeW]

@Hinfty   I understand your point.  I'm not a Fitbit employee, so please don't shoot the messenger.  I've never used any other fitness trackers, so I don't know if other brands handle time changes differently.  But this is how Fitbit handles time zone changes.

[Hinfty]

Got it. I think your fancy signature confused me here.

No worries and thanks for clearing up the (lousy) fitbit implementation.

I do recall that fitbit managed overflying several timezones (+7h) and the resulting shifts quite well, so I didn't expect that it would have such issues with a regular change.