Open devjolt opened 2 years ago
I am also having this issue.
In industrial setting, a boolean value would indicate the status of a valve changes between 0 and 1
Therefore a new point point indicates a change in a value
This is true for most IoT cases (data is only updated on significant change)
Therefore, an engineer troubleshooting an issue would expect that the rising or falling edge on the chart indicates the time of the change.
It doesn't.
Influx have not implemented this interpolation = step in a way that's consistent with an engineering understanding of timeseries data.
Is each step rendered half way between the two changing data points?
In other words, if: 12:00:00 - True 12:00:10 - False
12:00:05 - Datapoint rendered?
That seems to hold true for all datapoints on this chart...
This is especially an issue in histogram plot type where it is more misleading.
Is this ever going to be fixed?
Steps to reproduce:
from(bucket: "my_bucket") |> range(start: v.timeRangeStart, stop: v.timeRangeStop) |> filter(fn: (r) => r["host"] == "thing1") |> filter(fn: (r) => r["_field"] == "valve1") |> toFloat()
<-- to change data into 1 or 0!Expected behavior: See data charted with digital steps occurring where valve1 moves from true to false or vice versa (as with linear or smooth interpolation)
Actual behavior: Data charted, but steps do not align with data occurrence times (see image) NOTE: Data aligns just fine with linear or smooth interpolation.
In this image, see where data point is vs where the steps are:
In this image, see where more datapoints are (couldn't screen grab all at the same time, so you get red arrows!)
Environment info: hosted by influx