MOI.AbstractNLPEvaluator as a function or constraint container

[matbesancon]

It is unclear from MOI.Test and from the doc whether the AbstractNLPEvaluator expression should contain the nonlinear expression or the non-linear constraint.
Given l <= g(x) <= u,
the oracle interface (Jacobian, Hessian) provides information on g at a given x. However, the two valid examples in Test/test_nonlinear.jl, HS071 and FeasibilitySenseEvaluator have constraint_expr return the constraint expression g(x) <= u.

Having the bounds seem non-relevant since the BlockData API includes the bounds:
MOI.NLPBlockData(MOI.NLPBoundsPair.(lb, ub), evaluator, true)

[odow]

constraint_expr returns the full expression:

MathOptInterface.jl/src/nlp.jl

Lines 225 to 236 in 4d0f52e

	"""
	constraint_expr(d::AbstractNLPEvaluator, i)
	Returns an expression graph for the ``i\\text{th}`` constraint in the same
	format as described above, with an additional comparison operator indicating
	the sense of and bounds on the constraint. The right-hand side of the comparison
	must be a constant; that is, `:(x[1]^3 <= 1)` is allowed, while
	`:(1 <= x[1]^3)` is not valid. Double-sided constraints are allowed, in which
	case both the lower bound and upper bounds should be constants; for example,
	`:(-1 <= cos(x[1]) + sin(x[2]) <= 1)` is valid.
	"""
	function constraint_expr end

It's something that in retrospect wasn't needed, but it's too late to change now.

[matbesancon]

OK I see, I had missed this docstring I was looking at the one of the types mostly. Do you prefer closing this or keeping it open for the NL redesign?

[odow]

Let's close it. It's already on my mental list of "things that could be done better"