Datum access

[lrennels]

#677

Currently we have implemented the methods and constructors needed to access parameters and variables in the top level composite components of a model with a call like

m[:top, :par_name

next steps are to add more comprehensive testing and think about corner cases like renaming. also, do we want to be able to use this syntax for lower levels i.e. if :B is a child composite component of :top but not in the comps_dict of m.mi, should we be able to do

m[:B, :par name]

[codecov]

Codecov Report

Merging #704 into master will increase coverage by 0.28%.
The diff coverage is 96.29%.

@@            Coverage Diff             @@
##           master     #704      +/-   ##
==========================================
+ Coverage   79.74%   80.02%   +0.28%     
==========================================
  Files          38       38              
  Lines        2839     2854      +15     
==========================================
+ Hits         2264     2284      +20     
+ Misses        575      570       -5     

Flag	Coverage Δ
#unittests	80.02% <96.29%> (+0.28%)	⬆️

Impacted Files	Coverage Δ
src/core/instances.jl	79.13% <90.00%> (+0.82%)	⬆️
src/core/build.jl	100.00% <100.00%> (+4.10%)	⬆️
src/core/types/instances.jl	71.18% <100.00%> (-0.95%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2b70c8f...8719ac0. Read the comment docs.

[corakingdon]

Could you also update the "docs/src/internals/strucutres_3_instances.md" file just to reflect the two new fields for CompositeComponentInstances?

[corakingdon]

Looks great! The one thing I'm wavering on now is whether parameters and variables are the best names for the new fields in the ComposteComponentInstance. Maybe it should be something like parameter_refs instead. But maybe we should just merge this now and then open a new issue to ask David's thoughts on that later?

[lrennels]

Hmm so would the idea of not naming them variable and parameter be that they are difference since they are immutable at this level and just references to lower components as opposed to unique ones?

[corakingdon]

yeah mostly the second point-- just that they don't actually store parameter/variable values, but references to them. minor point, but I feel like David would have a clear opinion on it so we could just revisit it later!