Wrong equation count with complex numbers

[massimo ceraolo]

After a check it seems to me that this is not a duplicate, but I'm not totally sure.
Consider the following code:

model NumOfEq
  function F
    input Complex Z1, Z2;
    output Complex Z3;
  algorithm
    Z3 := Z1 + Z2;
  end F;
  model M4x2
    parameter Complex Z1 = Complex(1, 1);
    parameter Complex Z2 = Z1;
    Complex Z3 = F(Z1, Z2);
  end M4x2;
  model M4x4
    parameter Complex Z1 = Complex(1, 1);
    Complex Z2 = Z1;
    Complex Z3 = F(Z1, Z2);
  end M4x4;
end NumOfEq;

I expect to have 4 equations and variables in model M4x4 and 2 equations and variables in model M2x2.
However OM reports in the second case 4 equations and 2 variables.

[Martin Sjölund]

The flattened code looks like this:

  Real Z3.re = 2.0 "Real part of complex number";
  Real Z3.im = 2.0 "Imaginary part of complex number";
equation
  Z3 = NumOfEq.F(Z1, Z2);

So it seems we both evaluate the binding and move it to the equation section, resulting in duplicates :(

[Francesco Casella]

@perost, can this be fixed easily in the old front-end? Otherwise, we can schedule this for 2.0.0

[Per Östlund]

Fixed in c93ff57.