On 28.02.2015, at 16:44 , Felix Baumann <[email protected]> wrote: > @hanno could you take a look at that data and see whether all timestamps (not > upload but capturing) are from 1970? > If they are not then their cause would be something different and they > haven't been uploaded by RK. > > Ichnaea shouldn't allow uploads with time stamps that are before 2012 or > something (service start) and do a queue to delete those.
What we actually do here is only allow timestamps from the last 60 days. If the timestamp is older than that, we just autocorrect this to be "today" instead, assuming the phone clock was bad. But then again we don't actually use these observation timestamps for anything right now. For all internal processing the upload time / database insertion time is used. I wasn't aware of a bad clock time affecting the GPS readings that much, but it makes sense. @Djfe can you open a mozstumbler issue with your ideas stumbler side local time checks? Cheers, Hanno _______________________________________________ dev-geolocation mailing list [email protected] https://lists.mozilla.org/listinfo/dev-geolocation
