ModelLinkage: multi purpose method of linking local and remote data

[rburghol]

Tasks

• [ ] Can/should we use SCHEMATIC or MASSLINK for model data passing like in the old framework?
• [ ] What are requirements for setting data?
• [ ] Where in the hsp2 code is data from these passed and set?
• [ ] Final Data Model

Overview

These are connections

• See code: https://github.com/HARPgroup/HSPsquared/blob/specom/HSP2/om_model_linkage.py
• remote connections can be context sensitive. That is, in a sequential simulation (like standard HSPF) anything outside a given rchres is remote, but in a simultaneous solution they might not be.
  • remote: data would be obtained from the timeseries /RESULTS/xxxx
  • local: data would be obtained from the state /STATE/xxxx
  • maybe the decision on local/remote can be made either at declaration time, or, by the module invoking specl, so it automatically looks for state or ts depending?
  • can we include the path to model attribute in hdf5, then let the ts/state decision be made via code?

OM Inputs (old model)

• OM inputs: these are not an object class, but rather they are a set of stored object references, which we will likely replace with references to the global data store/dictionary
  • Old model used addInput() function, and inputs[] array to store links to other object data (read only). How to replicate this?

Data Model

• consider ts['/links/table'] like from_node, to_node, src_prop, dest_prop
  • Better to set these as /objects/system/...
• A water facility could store it's state table data in /objects/system/[hydroid]/...,