Fix initialisation freezing on empty primary data.

[staudtMarius]

Resolves #981

[sebastian-peter]

I adapted the code a tiny bit. Would be great to have tests for this as well, though, I think.

[danielfeismann]

Thank you for taking care of this!

One point to discuss, based on the data in vn_simona: Asuming I would increase the simulation start time by one hour from '2011-01-01T00:00:00Z' to 2011-01-01T01:00:00Z, this would cause the exception because time series 8c04e94e-76b0-4369-a55c-f5e1117fb83e has no entry for 01:00:00. But with our general approach we define values as valid until a new one at a certain tick arrise. In my humble opinion this case would be vaild, while some earlier start date (e.g. 2010-12-31T23:00:00Z') wouldn't.

[sebastian-peter]

One point to discuss, based on the data in vn_simona: Asuming I would increase the simulation start time by one hour from '2011-01-01T00:00:00Z' to 2011-01-01T01:00:00Z, this would cause the exception because time series 8c04e94e-76b0-4369-a55c-f5e1117fb83e has no entry for 01:00:00. But with our general approach we define values as valid until a new one at a certain tick arrise. In my humble opinion this case would be vaild, while some earlier start date (e.g. 2010-12-31T23:00:00Z') wouldn't.

That's a neat idea, however this sounds more complicated and could rather be handled in a new PR? For now I'm quite happy that there's an error message at all

[danielfeismann]

One point to discuss, based on the data in vn_simona: Asuming I would increase the simulation start time by one hour from '2011-01-01T00:00:00Z' to 2011-01-01T01:00:00Z, this would cause the exception because time series 8c04e94e-76b0-4369-a55c-f5e1117fb83e has no entry for 01:00:00. But with our general approach we define values as valid until a new one at a certain tick arrise. In my humble opinion this case would be vaild, while some earlier start date (e.g. 2010-12-31T23:00:00Z') wouldn't.

That's a neat idea, however this sounds more complicated and could rather be handled in a new PR? For now I'm quite happy that there's an error message at all

Fair point. Yes we should handle that in a new PR

[danielfeismann]

Just a question I got into my mind

[sebastian-peter]

I fully agree with @danielfeismann that this solution is far from ideal, but at least we display an error message now.

We should work on a more flexible solution that just takes the last value before simulation start if no value for tick 0 is present, as Daniel suggested.

One could also talk about sending out some sort of 0 kW data if the time series starts after simulation start, however, this could lead to strange agent behavior that is difficult to detect. It's probably better to require the user to explicitly set the value for tick 0.

[sebastian-peter]

Thanks a lot, @staudtMarius, for fixing this!