Illegal call to zeros with negative argument

[maltelenz]

The call to zeros(n_q - 1) here:

PowerSystems/PowerSystems/Examples/AC3ph/Precalculation.mo

Line 1142 in 69fd1e7

(zr_q, zx_q) := z_matrix(n_q, x_q, xsig_s, r_s, zeros(n_q-1), xsig_rq, r_rq);

seems to come down to zeros(-1), if we resolve n_q:

PowerSystems/PowerSystems/Examples/AC3ph/Precalculation.mo

Line 1137 in 69fd1e7

final parameter Integer n_q=size(xsig_rq,1);

PowerSystems/PowerSystems/Examples/AC3ph/Precalculation.mo

Line 1116 in 69fd1e7

parameter SIpu.Reactance[:] xsig_rq=fill(0,0)

giving the expression zeros(size(fill(0, 0), 1) - 1) ==> zeros(-1), which is illegal (all arguments to zeros need to be >=0), or am I missing something?

[rfranke]

Just checked with OpenModelica. It generates code with zeros(0), i.e. silently corrects this during model translation. The called function doesn't use the array for n < 2.

function z_fromEqCirc "Calculates impedance matrix z from equivalent circuit"
  extends Modelica.Icons.Function;
  input Integer n "transient order";
...
  input SIpu.Reactance[n-1] xm2_n(each unit="1") "coupling reactance";
...
  output SIpu.Reactance[n+1,n+1] zx(each unit="1")  "impedance matrix reactive";

algorithm 
...
    for k in 2:n loop
      zx[1:k,1:k] := zx[1:k,1:k] + fill(xm2_n[k-1], k, k);
    end for;
...
end z_fromEqCirc;

[maltelenz]

It is explicitly illegal: https://specification.modelica.org/maint/3.6/arrays.html#modelica:zeros

Wolfram System Modeler will reject models with such illegal uses.