Improves getting time step from Simulation type

[suyashbire1]

Fixes #648

[codecov]

Codecov Report

Merging #649 into master will decrease coverage by 0.03%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master     #649      +/-   ##
==========================================
- Coverage   78.03%   78.00%   -0.04%     
==========================================
  Files         118      118              
  Lines        2363     2364       +1     
==========================================
  Hits         1844     1844              
- Misses        519      520       +1     

---

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 2ae1cbb...ad2e6ca. Read the comment docs.

[glwagner]

Looks good, though there should be a docstring if this is user-facing.

Another possibility is to use Val to overload getproperty as we do for boundary condition directions:

https://github.com/climate-machine/Oceananigans.jl/blob/2ae1cbb86f484420dd2233745a5928a00bf6f77d/src/BoundaryConditions/coordinate_boundary_conditions.jl#L31

[glwagner]

I changed the title --- julia has types, but does not have classes.

[suyashbire1]

Another possibility is to use Val to overload getproperty as we do for boundary condition directions:

There are instances in the code where simulation.Δt is supposed to return a wizard. I guess overloading getproperty might get in the way of that code? @ali-ramadhan ?

[ali-ramadhan]

Yeah this is nice but ideally users should be able to just do simulation.Δt and get the current time step. Unfortunately get_Δt seems necessary and feels like a Java style getter-setter.

We do need in a few places:
https://github.com/climate-machine/Oceananigans.jl/blob/2ae1cbb86f484420dd2233745a5928a00bf6f77d/src/Simulations.jl#L156
https://github.com/climate-machine/Oceananigans.jl/blob/2ae1cbb86f484420dd2233745a5928a00bf6f77d/src/Simulations.jl#L164

I think we should merge this PR but maybe figure out if there's an easier way to interact with simulations. Printing progress messages seems particularly cumbersome as you end up having to dive into multiple layers of types to get what you want (as shown by the example in #648).