I am trying to figure out how influx is calculating the times returned from a flux query with an aggregateWindow.
For example: I am doing a pretty basic aggregateWindow(every: 300d, fn: mean). If I use a range with start 2019-03-30T14:35:34.932Z the first entry in the returned data has a time of 2019-04-14T00:00:00Z
It doesn't depend on the data. Even if I change to a year where there is no data in the first period it still return the same time (in that year, i.e. 2022-04-14).
I can't find anything in the docs about this apart from a hint regarding using weeks "Flux increments weeks from the Unix epoch, which was a Thursday.". Does this apply to all values "every"? I.e. are all windows being aligned to the epoch?
How would I go about implementing something similar in Java?