Issue with linspace() not evaluated in NF

[Francesco Casella]

Please check the ​ScalableTestSuite.Thermal.DistrictHeating.ScaledExperiments.HeatingSystem_N_5 model. It fails with this error:

ScalableTestSuite_ScalableTestSuite.Thermal.DistrictHeating.ScaledExperiments.HeatingSystem_N_5_08bnd.c:167:60: error: passing 'int' to parameter of incompatible type 'real_array_t' (aka 'struct base_array_s')
  data->simulationInfo->realParameter[5] = (1.0 + real_get(linspace(0.0, 1.348, ((modelica_integer) 5)), 4)) * (10000000.0);
                                                           ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The NF produces this flat declaration

parameter Real Cu[1](quantity = "HeatCapacity", unit = "J/K") =
  (1.0 + linspace(0.0, 1.348, 5)[1]) * 10000000.0 "Heat capacity of heated units";

while the old FE gives

parameter Real Cu[1](quantity = "HeatCapacity", unit = "J/K") =
   10000000.0 "Heat capacity of heated units";

I'm not sure if the code generation should be extended to handle the raw expression involving linspace(), or if the NF should rather evaluate the constant expression and just pass that to the backend, as the old FE did.

Possibly related to #4861

[Martin Sjölund]

OMC actually doesn't know anything about linspace except the definition (which uses inlining instead of a builtin function). The new FE needs to do the same or introduce handling of linspace everywhere.

[Francesco Casella]

Fixed sometime on May 8th, see ​simulation report.