Breaking changes for MOI v0.10

[metab0t]

Hope CI is happy this time.

[odow]

Yeah it should run now.

[codecov]

Codecov Report

Merging #147 (f39c58e) into master (6a6cdbf) will decrease coverage by 0.18%.
The diff coverage is 83.73%.

@@            Coverage Diff             @@
##           master     #147      +/-   ##
==========================================
- Coverage   54.65%   54.46%   -0.19%     
==========================================
  Files           9        9              
  Lines        4007     4030      +23     
==========================================
+ Hits         2190     2195       +5     
- Misses       1817     1835      +18     

Impacted Files	Coverage Δ
src/MOI/MOI_wrapper.jl	86.69% <83.73%> (-0.77%)	⬇️
src/helper.jl	41.13% <0.00%> (-0.64%)	⬇️

---

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 6a6cdbf...f39c58e. Read the comment docs.

[odow]

Nice! Things are working at least. There are still a few tests being excluded, but it might be easiest to merge this PR first, and then tackle them individually.

[metab0t]

The callback tests now all pass.
I use Revise.jl and forget to restart Julia yesterday :(

[metab0t]

Conflict tests and farkas tests that exist in MOI.Test are removed.

[metab0t]

The MOI test config is similar with Gurobi.jl. The conic tests throw error if MOI.ConstraintDual and MOI.DualObjectiveValue are not excluded.

[metab0t]

I have a look at MOI.Test.test_solve_conflict_EqualTo.
The IIS obtained by Xpress says that b2 is not in conflict.

[joaquimg]

I have a look at MOI.Test.test_solve_conflict_EqualTo.
The IIS obtained by Xpress says that b2 is not in conflict.

Xpress IIS does not track variable bounds

[joaquimg]

@metab0t do you have plans to look into the failing tests?

[metab0t]

@joaquimg Yes, I plan to look at these problems.
I will look at these failing tests today.

"test_objective_set_via_modify"
"test_model_ListOfConstraintAttributesSet"
"test_objective_get_ObjectiveFunction_ScalarAffineFunction"

[metab0t]

@joaquimg
For test_objective_set_via_modify the error happens here
We need to change model.objective_type in

function MOI.modify(
    model::Optimizer,
    c::MOI.ConstraintIndex{MOI.ScalarAffineFunction{Float64}, <:Any},
    chg::MOI.ScalarCoefficientChange{Float64}
)

I cannot think of a good solution now.
Another error happens in line 514, the empty model should have zero attributes when calling MOI.get(model, MOI.ListOfModelAttributesSet()) and I have fixed it.

[metab0t]

@joaquimg
For test_objective_get_ObjectiveFunction_ScalarAffineFunction, there is a corner case in Xpress.jl.
If the MOI.ObjectiveFunction is set but MOI.ObjectiveSense is not set, then model.is_feasibility=true and MOI.get(model::Optimizer, ::MOI.ObjectiveFunctionType) = nothing.
Should we add another field to indicate whether objective function is set because is_feasibility has been used to indicate whether objective sense is set?

[odow]

Take a look at GLPK etc.
https://github.com/jump-dev/GLPK.jl/blob/d11ae27aaa3666217786850c564aeffa138a6dac/src/MOI_wrapper/MOI_wrapper.jl#L88-L89

I switched to having a objective_sense::Union{Nothing,MOI.OptimizationSense}.

[metab0t]

Take a look at GLPK etc. https://github.com/jump-dev/GLPK.jl/blob/d11ae27aaa3666217786850c564aeffa138a6dac/src/MOI_wrapper/MOI_wrapper.jl#L88-L89

I switched to having a objective_sense::Union{Nothing,MOI.OptimizationSense}.

I have made similar changes in latest commit.

[metab0t]

@joaquimg Any plans to merge this PR and prepare a version bump?

[joaquimg]

While this was opened there were a few fixes being applied in other PRs. I was waiting for them to complete so that we don't need to backport.
If nothing else show up this week I will proceed with the merge.
Thanks!

[joaquimg]

Still requires:
jump-dev/MathOptInterface.jl#1648

[joaquimg]

Now waiting for:
jump-dev/MathOptInterface.jl#1650