Error messages with integer constraint vectors

[baggepinnen]

The following throws an error, reproduced below, due to lb = fill(-10, length(params)) being a vector of Ints instead of floats. Could automatic conversion be applied, alternatively, a better error message be thrown? The only clue as to what is wrong is in MathOptInterface.GreaterThan{Int64}, but this is still far from obvious

using Ipopt, OptimizationMOI, Optimization
params = randn(2)
fopt = OptimizationFunction((x, _)->x'x)
prob = OptimizationProblem(fopt, params;
    lb = fill(-10, length(params)),
    ub = fill(10, length(params)),
)
solver = Ipopt.Optimizer()
res = solve(prob, solver)

julia> res = solve(prob, solver)
ERROR: MathOptInterface.UnsupportedConstraint{MathOptInterface.VariableIndex, MathOptInterface.GreaterThan{Int64}}: `MathOptInterface.VariableIndex`-in-`MathOptInterface.GreaterThan{Int64}` constraint is not supported by the model.
Stacktrace:
  [1] correct_throw_add_constraint_error_fallback(model::Ipopt.Optimizer, func::MathOptInterface.VariableIndex, set::MathOptInterface.GreaterThan{Int64}; error_if_supported::MathOptInterface.AddConstraintNotAllowed{MathOptInterface.VariableIndex, MathOptInterface.GreaterThan{Int64}})
    @ MathOptInterface ~/.julia/packages/MathOptInterface/8f6oN/src/constraints.jl:211
  [2] correct_throw_add_constraint_error_fallback(model::Ipopt.Optimizer, func::MathOptInterface.VariableIndex, set::MathOptInterface.GreaterThan{Int64})
    @ MathOptInterface ~/.julia/packages/MathOptInterface/8f6oN/src/constraints.jl:202
  [3] throw_add_constraint_error_fallback(model::Ipopt.Optimizer, func::MathOptInterface.VariableIndex, set::MathOptInterface.GreaterThan{Int64}; kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
    @ MathOptInterface ~/.julia/packages/MathOptInterface/8f6oN/src/constraints.jl:160
  [4] throw_add_constraint_error_fallback(model::Ipopt.Optimizer, func::MathOptInterface.VariableIndex, set::MathOptInterface.GreaterThan{Int64})
    @ MathOptInterface ~/.julia/packages/MathOptInterface/8f6oN/src/constraints.jl:154
  [5] add_constraint(model::Ipopt.Optimizer, func::MathOptInterface.VariableIndex, set::MathOptInterface.GreaterThan{Int64})
    @ MathOptInterface ~/.julia/packages/MathOptInterface/8f6oN/src/constraints.jl:148
  [6] __solve(prob::OptimizationProblem{true, OptimizationFunction{true, SciMLBase.NoAD, var"#25#26", Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED_NO_TIME), Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing}, Vector{Float64}, SciMLBase.NullParameters, Vector{Int64}, Vector{Int64}, Nothing, Nothing, Nothing, Nothing, Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}}}, opt::Ipopt.Optimizer; maxiters::Nothing, maxtime::Nothing, abstol::Nothing, reltol::Nothing, kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
    @ OptimizationMOI ~/.julia/packages/OptimizationMOI/08JM7/src/OptimizationMOI.jl:358
  [7] __solve

[Vaibhavdixit02]

Yeah I think we can do the type conversion, improving the error is probably not possible because this is at the MOI level but we can do the conversion before passing it on