Originally from Daniel Thom, on #565:
About 1 (2?) year ago you, Gabriel, and I discussed an idea: parameterize the time series type by the type of its values, e.g. SingleTimeSeries{Float64} (analogous to Vector{Float64}). Is now the time to do it? It would solve this problem in a better way (the time series key has the type).
I opened #566 in part due to the above comment: that's a snapshot of the current situation in IS4 + PSY6, of places where we've added parallel structs for time-varying-ness. The above might allow for some simplifications: e.g. TimeSeriesFunctionData{T} could become SingleTimeSeries{T <: FunctionData}.
edit: also relevant, PSY #1150
Originally from Daniel Thom, on #565:
I opened #566 in part due to the above comment: that's a snapshot of the current situation in IS4 + PSY6, of places where we've added parallel structs for time-varying-ness. The above might allow for some simplifications: e.g.
TimeSeriesFunctionData{T}could becomeSingleTimeSeries{T <: FunctionData}.edit: also relevant, PSY #1150