Today I pushed a number of updates to address Issue #795, a serious weakness in tariff evaluation. The problem was that the process of costing a usage profile assumed that all profiles started in the next timeslot, but in fact it turns out that only the factored-customer models generate profiles in this way. After considerable discussion on how to address this, I have implemented a set of changes that allow customer models to communicate the start time of their profiles to the evaluator.
These changes affect common, server-interface, and all the customer models. If you are working in a source environment, I invite you to pull these changes and try them out. After a bit more testing, and assuming that no bad behaviors result, I intend to deploy a new snapshot on Monday or Tuesday. Note that the changes in common are very unlikely to affect brokers unless your broker is using the TariffEvaluationHelper to inspect tariffs.
There remains a small problem in the EV model, Issue #814. The symptom is that we get occasional errors in which the EV customers are unable to regulate by the requested amount. Govert and I hope to have this resolved very quickly.