Issue with each in start attribute binding with the NF

[Francesco Casella]

Please check ​Modelica.Electrical.Digital.Examples.DFFREG. The compiler reports

[/var/lib/hudson/slave/workspace/OpenModelica_TEST_LIBS/OpenModelica/OMCompiler/build/lib/omlibrary/Modelica 3.2.2/Electrical/Digital.mo:682:63-682:73:writable] Error: Type mismatch in binding fixed = true, expected subtype of Boolean[2, 2], got type Boolean.

From what I understand looking at the model, fixed = true needs to be applied to an array of 2 elements (not 2x2) by means of a scalar 'each' value.

[Francesco Casella]

As of yesterday, the error was

Type mismatch in binding fixed = true, expected subtype of Boolean[2], got type Boolean

which is better (the expected type is correct) but not yet there (some problems with each).

In fact, I'm not sure the MSL code is correct here:

delay(inertialDelaySensitive(each y(start=L.'U', fixed=true))),

shouldn't it rather be

delay(inertialDelaySensitive(y(each start=L.'U', each fixed=true))),

[Per Östlund]

Replying to casella:

In fact, I'm not sure the MSL code is correct here:

delay(inertialDelaySensitive(each y(start=L.'U', fixed=true))),

shouldn't it rather be

delay(inertialDelaySensitive(y(each start=L.'U', each fixed=true))),

No, the first is correct since inertialDelaySensitive is the array in this case. But yeah, it's not always clear how each should be used. Both the old and new instantiation is a bit confused about each actually, and the whole handling of each probably needs to be redesigned a bit in the new instantiation.

[Martin Sjölund]

Smaller model for this that triggers NF (and also the old one):

model O
  Real n[2];
end O;

model M
  O o(each n(fixed=true));
end M;

[Francesco Casella]

This model now passess the FE and breaks much later in the backend (see #4817). I'm not sure if this issue has been solved thoroughly.