Codecov Report

Merging #219 (d23938f) into master (e31b30c) will decrease coverage by 2.04%.
The diff coverage is 91.44%.

❗ Current head d23938f differs from pull request most recent head 7f9984c. Consider uploading reports for the commit 7f9984c to get more accurate results

@@            Coverage Diff             @@
##           master     #219      +/-   ##
==========================================
- Coverage   87.32%   85.27%   -2.05%     
==========================================
  Files           8        7       -1     
  Lines         418      292     -126     
==========================================
- Hits          365      249     -116     
+ Misses         53       43      -10     

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 e31b30c...7f9984c. Read the comment docs.

Okay, after this experience, I think I'm now on board with:

1. The required bridges thing. It made SCS much simpler than the direction I was originally going down.
2. The OptimizerCache stuff. Not necessarily as a replacement layer, but as a simpler way of converting into matrix-type formats
3. The one-shot optimize!(dest, src).

All-in-all, this rewrite of SCS.jl was not a fun experience, but I think we're in a much better place than we were.

@blegat did you actually try jump-dev/MathOptInterface.jl#1592 on a solver? It isn't working.

empty! is not reaching the inner optimizer

julia> model = MOI.Bridges.full_bridge_optimizer(
           MOI.Utilities.CachingOptimizer(
               MOI.Utilities.UniversalFallback(MOI.Utilities.Model{Float64}()),
               SCS.Optimizer(),
           ),
           Float64,
       );

julia> MOI.set(model, MOI.Silent(), true)

julia> x, ci = MOI.add_constrained_variables(model, MOI.Nonnegatives(2));

julia> MOI.optimize!(model)

julia> MOI.empty!(model)

julia> MOI.get(model, MOI.TerminationStatus())
OPTIMAL::TerminationStatusCode = 1