Cbc backend returns primal status feasible for infeasible problem

[DrChainsaw]

Sorry if this is the wrong place for this.

MWE with a constraint that 10 binary variables shall add up to 11:

julia> using JuMP, Cbc

julia> mmm = Model(with_optimizer(Cbc.Optimizer));

julia> @variable(mmm, x[1:10], Bin);

julia> @objective(mmm, Max, sum(x))
x[1] + x[2] + x[3] + x[4] + x[5] + x[6] + x[7] + x[8] + x[9] + x[10]

julia> @constraint(mmm, infeasible, sum(x) == 11)
infeasible : x[1] + x[2] + x[3] + x[4] + x[5] + x[6] + x[7] + x[8] + x[9] + x[10] == 11.0

julia> JuMP.optimize!(mmm)
Welcome to the CBC MILP Solver
Version: 2.9.9
Build Date: Dec 31 2018

command line - Cbc_C_Interface -solve -quit (default strategy 1)
Problem is infeasible - 0.00 seconds
Total time (CPU seconds):       0.01   (Wallclock seconds):       0.01

julia> JuMP.termination_status(mmm)
INFEASIBLE::TerminationStatusCode = 2

julia> JuMP.primal_status(mmm)
FEASIBLE_POINT::ResultStatusCode = 1

Also sorry if this is intended/unavoidable behaviour or if I'm just misinformed (not very experienced with optimization).

For the real problem I am interested in a feasible solution even though it is not optimal so I ran into the issue because I only checked the primal status.

I guess it is enough to look at the termination status to determine that the solver considered the problem infeasible and only look at the primal status if not infeasible.

[mlubin]

This indeed appears to be a bug.

[blegat]

The if and elseif might need to be swapped here: https://github.com/JuliaOpt/Cbc.jl/blob/17e0bd7b1e7f309eebf0426a00cfad7153cf3bf3/src/MOI_wrapper.jl#L536-L540