rl-institut
Time parameter input should be rethought
With the changes proposed in #625, one need the time parameters of the simulation_settings in order to be able to parse a json file and add timeindex to the series present in the json file. We should decide on a fixed format for input of these settings and we should remove them from the json to python types parser
Can you elaborate which of the simulation_settings we need to rethink or define better?
We need start_date, evaluated_period and timesteps (or periods, I find periods conceptually wrongly named and not super useful as input)as parameters in order to compute the DatetimeIndex which we want to append to all the series provided by EPA when while parse their json file as input to MVS (there are implicitly timeseries, but for MVS we need to turn the implicit into explicit ;)).
We need to unambiguously set the format of start_date (either numbers of seconds since epoch (as int) or iso timestamp), EPA provide us with "number of seconds" format.
Can the unit of timestep be something else than minutes, if not we just need to make sure the user did not input "hours" for example (or we could allow minutes or hours or days?). Same questions for evaluated_period.
We also need to consider what we want to happen when we convert back from mvs nested dict_values to json.
The TODO in tests/test_F0_output.py should be addressed when addressing this issue, it was introduced in #625
start_date-> human readable ISO (2020-11-05 00:00:00)
evaluated_period -> 1...365 days
timesteps -> nicht periods -> nur in minuten anzugeben (grade gehen ausschließlich 60 min).
Communicate this to EPA.
EPA: They should be able to store timeseries of 365 days, eventhough a user might be inclined to start simulating with only 7 days. -> create issue on gitlab?