I've been passing milliseconds to PipeHandler, entries decode properly, but CPR never resolves to latitude/longitude since the window is only 10 milliseconds (not 10 seconds). The tests use seconds (and time.time()), but it seems that the ways CPR doesn't resolve don't surface easily- no reference= and the timestamp units.
The best implications that ts must be seconds:
- the tests, which explicitly test with seconds
- the _.*_tolerance values, which imply their values are in X per second.
- the
timestamp (with no units specified), combined with pair_window or evicition_ttl (which are specified as being seconds)
I could add a quick docfix for it, but ideally completely absurd values should be rejected- if millis are passed in the year is approximately 2533. Some cutoff would quickly surface the problem with the value.
This is almost but not fully related to #200.
I've been passing milliseconds to PipeHandler, entries decode properly, but CPR never resolves to latitude/longitude since the window is only 10 milliseconds (not 10 seconds). The tests use seconds (and time.time()), but it seems that the ways CPR doesn't resolve don't surface easily- no
reference=and the timestamp units.The best implications that ts must be seconds:
timestamp(with no units specified), combined withpair_windoworevicition_ttl(which are specified as being seconds)I could add a quick docfix for it, but ideally completely absurd values should be rejected- if millis are passed in the year is approximately 2533. Some cutoff would quickly surface the problem with the value.
This is almost but not fully related to #200.