Oscar Dowson

CPLEX only supports `relative_gap` if the problem is a MIP.

I guess we could do something like ```Julia function MOI.get(model::Optimizer, attr::MOI.RelativeGap) _throw_if_optimize_in_progress(model, attr) p = Ref{Cdouble}() ret = CPXgetmiprelgap(model.env, model.lp, p) if ret == CPXERR_NOT_MIP return NaN end _check_ret(model, ret)...

Actually, the correct solution is to throw the new `MOI. GetAttributeNotAllowed` error (to be released in MOI v1.5.0), with the message that the problem is not a mixed-integer problem. x-ref:...

Hmm. It isn't really intended as an input-output layer, it's really just to help internal solver APIs. We should finish tidying up MatrixOptInterface.jl?

I guess a tutorial explaining how to get the LP standard form would be useful: https://github.com/jump-dev/MatrixOptInterface.jl

I started looking into MatrixOptInterface for this, but it really has the opposite goal: converting from matrix formulations into MOI. I think we really need something in MOI that converts...

Trying to handle all combinations that people might ask for just seems like more trouble than it's worth. So too supporting conic and quadratic matrix problems.

A bridge would be good. Then solvers could, if they choose, drop support for `ScalarAffineFunction` and use this one instead. If it turns out to be much better, then we...

Presumably it's the use of `::Integer` in `src/functions.jl` which conflicts with the MOI set `MOI.Integer`. Perhaps just use `::Core.Integer` instead?

The bridge errors are just printing. More serious is this ```Julia Testing /home/runner/work/MathOptInterface.jl/MathOptInterface.jl/test/Bridges/Constraint/slack.jl 311 Scalar slack: Error During Test at /home/runner/work/MathOptInterface.jl/MathOptInterface.jl/test/Bridges/Constraint/slack.jl:14 312 Got exception outside of a @test 313 UndefVarError:...