Record constructors are wrongly re-used by the NF

[Francesco Casella]

Please check ​Modelica.Media.Examples.MixtureGases. The NF gives this error:

[/var/lib/hudson/slave/workspace/OpenModelica_TEST_LIBS/OpenModelica/OMCompiler/build/lib/omlibrary/Modelica trunk/Media/IdealGases/Common/package.mo:835:5-836:71:writable] 
Error: Cannot resolve type of expression array(Modelica.Media.IdealGases.Common.Functions.cp_T({Modelica.Media.IdealGases.Common.DataRecord("CH4", 0.01604246, -4650159.63885838, 624355.7409524474, 1000.0, {-176685.0998, 2786.18102, -12.0257785, 0.0391761929, -3.61905443e-05, 2.026853043e-08, -4.976705489999999e-12}, {-23313.1436, 89.0432275}, {3730042.76, -13835.01485, 20.49107091, -0.001961974759, 4.72731304e-07, -3.72881469e-11, 1.623737207e-15}, {75320.6691, -121.9124889}, 518.2791167938085), Modelica.Media.IdealGases.Common.DataRecord("C2H6", 0.03006904, -2788633.890539904, 395476.3437741943, 1000.0, {-186204.4161, 3406.19186, -19.51705092, 0.0756583559, -8.204173219999999e-05, 5.0611358e-08, -1.319281992e-11}, {-27029.3289, 129.8140496}, {5025782.13, -20330.22397, 33.2255293, -0.00383670341, 7.23840586e-07, -7.3191825e-11, 3.065468699e-15}, {111596.395, -203.9410584}, 276.5127187299628), Modelica.Media.IdealGases.Common.DataRecord("C3H8", 0.04409562, -2373931.923397381, 334301.1845620949, 1000.0, {-243314.4337, 4656.27081, -29.39466091, 0.1188952745, -0.0001376308269, 8.814823909999999e-08, -2.342987994e-11}, {-35403.3527, 184.1749277}, {6420731.680000001, -26597.91134, 45.3435684, -0.00502066392, 9.471216939999999e-07, -9.57540523e-11, 4.00967288e-15}, {145558.2459, -281.8374734}, 188.5555073270316), Modelica.Media.IdealGases.Common.DataRecord("C4H10_n_butane", 0.0581222, -2164233.28779709, 330832.0228759407, 1000.0, {-317587.254, 6176.331819999999, -38.9156212, 0.1584654284, -0.0001860050159, 1.199676349e-07, -3.20167055e-11}, {-45403.63390000001, 237.9488665}, {7682322.45, -32560.5151, 57.3673275, -0.00619791681, 1.180186048e-06, -1.221893698e-10, 5.250635250000001e-15}, {177452.656, -358.791876}, 143.0515706563069), Modelica.Media.IdealGases.Common.DataRecord("N2", 0.0280134, 0.0, 309498.4543111511, 1000.0, {22103.71497, -381.846182, 6.08273836, -0.00853091441, 1.384646189e-05, -9.62579362e-09, 2.519705809e-12}, {710.846086, -10.76003744}, {587712.406, -2239.249073, 6.06694922, -0.00061396855, 1.491806679e-07, -1.923105485e-11, 1.061954386e-15}, {12832.10415, -15.86640027}, 296.8033869505308), Modelica.Media.IdealGases.Common.DataRecord("CO2", 0.0440095, -8941478.544405185, 212805.6215135368, 1000.0, {49436.5054, -626.411601, 5.30172524, 0.002503813816, -2.127308728e-07, -7.68998878e-10, 2.849677801e-13}, {-45281.9846, -7.04827944}, {117696.2419, -1788.791477, 8.291523189999999, -9.22315678e-05, 4.86367688e-09, -1.891053312e-12, 6.330036589999999e-16}, {-39083.5059, -26.52669281}, 188.9244822140674)}[i], state.T) for i in 1:6) * (state.X). 
The operands have types Real[6], Real[2] in component <NO_COMPONENT>.

The model contains two medium package declaration: Medium1, with 2 components, and Medium2, with 6 components. Apparently, the wrong cp_T function is called in the offending statement.

Maybe related to #5075?

[Francesco Casella]

The error is now in NFTyping.typeClassSections

[Modelica trunk/Media/IdealGases/Common/package.mo:714:9-716:97:writable]
Error: Type mismatch for positional argument 3 in Modelica.Media.Examples.MixtureGases.Medium2.ThermodynamicState(
X={Modelica.Media.Common.smoothStep(x, state_a.X[$i1], state_b.X[$i1], x_small) for $i1 in 1:2}).
The argument has type:
  Real[2]
expected type:
  Real[6]

the reason appears to be similar.

[Francesco Casella]

See also #5017

[Francesco Casella]

As noted by @perost in ticket:5075#comment:7 the issue is that

The whole handling of functions and record constructors needs to be redesigned, because currently they're shared in many places where they shouldn't be shared. That's why you get missing constructors sometimes, because the wrong constructors are reused and the correct ones are newer generated.

I closed #5075 as a duplicate and just kept this one, because I think this test case shows what the problem is in the simplest possible way.

[Francesco Casella]

This issue is now resolved. The model still fails, but for some other reason.