Manage time intervals for minimum metadata dictionary
Summary
Deal with current warnings of time metadata not being accessed.
Current behaviour/setbacks
Currently running the pipeline (v 0.1.51) results in this warning being printed repeatedly:
2023-01-11T16:19:14+0000 - WARNING:Signal: Signal: Unable to convert columns to minutes: "Can't open attribute (can't locate attribute: 'time_settings/timeinterval')"
Additionally, using the outputs from the pipeline, invoking a Signal
object or its derivatives prints:
Warning:Signal: Unable to convert columns to minutes: "Can't open attribute (can't locate attribute: 'time_settings/timeinterval')"
And the pandas.DataFrame
objects produced have columns as serial integers from 0, 1, 2 ... the end.
This doesn't completely block analysis, but would break visualisation/plotting code that assumes that the columns are minutes (in particular, postprocessor.routines
).
Desired behaviour/advantages
We need to come up with a way to a) silence the warning during the run of aliby pipeline or b) Include timeinterval in the minimal default metadata.
Ideally, signal columns should use minutes, e.g. 0, 5, 10 ... the end (if the interval is 5 minutes).
Implementation sketch
The attribute mentioned in the warning may not have been passed correctly
Still unsure about implementation, but it is likely to require the removal of that warning in Signal.