cscotta / deschutes Goto Github PK

I've been messing around with this library (awesome stuff, btw) and I'm a little confused as to whether event timestamps should be in seconds or milliseconds.

You mention milliseconds here - https://github.com/cscotta/deschutes/blob/master/src/main/java/com/cscotta/deschutes/api/Event.java#L7

But then is the windowLength in the OLAPAggregator in seconds, milliseconds, or "positions"? - https://github.com/cscotta/deschutes/blob/master/src/main/java/com/cscotta/deschutes/aggregation/OLAPAggregator.java#L42

Because down here - https://github.com/cscotta/deschutes/blob/master/src/main/java/com/cscotta/deschutes/aggregation/OLAPAggregator.java#L79 we multiply the windowLength by 1000 which makes me wonder if expired variable contains all of the events outside of a second-resolution window.

But to determine if an event has expired, we're just subtracting windowLength from currentTs - https://github.com/cscotta/deschutes/blob/master/src/main/java/com/cscotta/deschutes/aggregation/OLAPAggregator.java#L79

Also, when you are generating sample data, you generate whatever EPS, but they all have the same second-level time resolution - https://github.com/cscotta/deschutes/blob/master/src/main/java/com/cscotta/deschutes/example/source/GenerateData.java#L45 and https://github.com/cscotta/deschutes/blob/master/src/main/java/com/cscotta/deschutes/example/source/SyntheticProtobufs.java#L34

I was wondering why that was the case.

Any help would be appreciated. Thanks again.