[FileFormats] support other number types other than Float64

[hdavid16]

Hi, I am trying to read this model file so that it uses Int in order to use it with MiniZinc/Chuffed. However, I get an error on the types. Is it possible to use MOI.read_from_file so that values are not converted from Int to Float?

feasibility.mof.json

m = MOI.Utilities.Model{Int}()
MOI.read_from_file(m, "feasibility.mof.json")

ERROR: UnsupportedConstraint: `MathOptInterface.ScalarAffineFunction{Float64}`-in-`MathOptInterface.EqualTo{Float64}` constraints are not supported by the
solver you have chosen, and we could not reformulate your model into a
form that is supported.

To fix this error you must choose a different solver.

pkg> st JuMP
Status `~/Documents/rai-supply-chain-demo/Project.toml`
  [4076af6c] JuMP v1.26.0

[odow]

This needs some features to be added to the MOI.FileFormats.MOF submodule. Currently it assumes Float64.

[chriscoey]

@odow relatedly, we also seemed to have trouble writing Int models to LP format. I didn't see Int tests in the LP format tests, and I see

MathOptInterface.jl/src/FileFormats/LP/LP.jl

Line 96 in 9aba539

model = Model{Float64}()

seems to hardcode Float64. Perhaps the title of this issue can be generalized, or a separate issue can be created?

[odow]

I've changed the title

[odow]

https://github.com/jump-dev/MathOptInterface.jl/actions/runs/15457235366

[odow]

With master, this now yields:

julia> m = MOI.Utilities.Model{Int}()
MOIU.Model{Int64}
├ ObjectiveSense: FEASIBILITY_SENSE
├ ObjectiveFunctionType: MOI.ScalarAffineFunction{Int64}
├ NumberOfVariables: 0
└ NumberOfConstraints: 0

julia> MOI.read_from_file(m, "feasibility.mof.json"; coefficient_type = Int)

julia> m
MOIU.Model{Int64}
├ ObjectiveSense: FEASIBILITY_SENSE
├ ObjectiveFunctionType: MOI.ScalarAffineFunction{Int64}
├ NumberOfVariables: 15
└ NumberOfConstraints: 34
  ├ MOI.ScalarAffineFunction{Int64} in MOI.EqualTo{Int64}: 4
  ├ MOI.ScalarAffineFunction{Int64} in MOI.GreaterThan{Int64}: 6
  ├ MOI.ScalarAffineFunction{Int64} in MOI.LessThan{Int64}: 9
  └ MOI.VariableIndex in MOI.ZeroOne: 15

julia> print(m)
Feasibility

Subject to:

ScalarAffineFunction{Int64}-in-EqualTo{Int64}
 (0) - (1) x_C + (1) x_S3 + (1) x_S2 == (0)  (c1)
 (0) - (1) x_D + (1) x_S5 + (1) x_S4 == (0)  (c2)
 (0) + (1) x_S1 - (1) x_A + (1) x_S0 == (0)  (c3)
 (0) + (1) x_A == (1)  (c4)

ScalarAffineFunction{Int64}-in-GreaterThan{Int64}
 (0) - (1) x_E + (1) x_S2 >= (0)  (c1_1)
 (0) - (1) x_E + (1) x_S3 >= (0)  (c2_1)
 (0) - (1) x_D + (1) x_S1 - (1) x_I - (1) x_C >= (-2)  (c3_1)
 (0) - (1) x_F - (1) x_G + (1) x_S4 >= (-1)  (c4_1)
 (0) + (1) x_S0 - (1) x_B >= (0)  (c5)
 (0) - (1) x_H + (1) x_S5 >= (0)  (c6)

ScalarAffineFunction{Int64}-in-LessThan{Int64}
 (0) + (1) x_S1 - (1) x_I <= (0)  (c1_2)
 (0) - (1) x_D + (1) x_S1 <= (0)  (c2_2)
 (0) - (1) x_G + (1) x_S4 <= (0)  (c3_2)
 (0) - (1) x_E + (1) x_S2 <= (0)  (c4_2)
 (0) + (1) x_S1 - (1) x_C <= (0)  (c5_1)
 (0) - (1) x_E + (1) x_S3 <= (0)  (c6_1)
 (0) - (1) x_F + (1) x_S4 <= (0)  (c7)
 (0) + (1) x_S0 - (1) x_B <= (0)  (c8)
 (0) - (1) x_H + (1) x_S5 <= (0)  (c9)

VariableIndex-in-ZeroOne
 x_D ∈ {0, 1}
 x_E ∈ {0, 1}
 x_S1 ∈ {0, 1}
 x_I ∈ {0, 1}
 x_H ∈ {0, 1}
 x_C ∈ {0, 1}
 x_S3 ∈ {0, 1}
 x_S5 ∈ {0, 1}
 x_S2 ∈ {0, 1}
 x_A ∈ {0, 1}
 x_F ∈ {0, 1}
 x_G ∈ {0, 1}
 x_S0 ∈ {0, 1}
 x_B ∈ {0, 1}
 x_S4 ∈ {0, 1}