Possibly missing diagnostics for vector of vector of variables

[LebedevRI]

using JuMP
model = Model()
@variable(model, x[1:10][1:5])

results in

5-element Vector{VariableRef}:
 x[1]
 x[2]
 x[3]
 x[4]
 x[5]

... so the outer dimension is "lost".
Should this be diagnosed?

Perhaps the intended behavior is

using JuMP
model = Model()
x = [@variable(model, [1:5], base_name="x[$i]") for i in 1:10]

I encountered this when creating a 3-dimensional variable, which is a vector of symmetrical 2D matrixes:

x = [@variable(model, [1:5,1:5], base_name="x[$f]", Symmetric, Bin) for f in 1:10]

[odow]

@variable(model, x[1:10][1:5])

This is not valid JuMP syntax.

We should throw an error instead of silently dropping the [1:10]

[odow]

Fix is #3756

This is the recommended syntax:

using JuMP
model = Model()
x = [@variable(model, [1:5], base_name="x[$i]") for i in 1:10]

[LebedevRI]

@odow thank you!