Missing each when setting start values for arrays with dialog

[AnHeuermann]

Description

When changing a start value of an array state variable with a dialog in diagram view in OMEdit and setting it to true: start value is used to initialize the resulting model is missing the each key word. Because of that the model can't be simulated.

Steps to Reproduce

Create some model with an array state variable:

model TestComp
  parameter Real a[2]={10,5};
  Real x[2](start={0,0}) annotation (Dialog(tab="Setting",showStartAttribute=true));
equation
  der(x[1]) = a[1]*x[1];
  der(x[2]) = a[2]*x[2];
end TestComp;

Drag the model TestComp into a new model mwe using the diagram view in OMEdit.

Then change the start value of x to {1,2} with the Element Parameter dialog and activate true: start-value is used to initialize.

The resulting model will look like:

model mwe
  TestComp testComp(x(fixed = true, start = {1, 2}))  annotation(
    Placement(visible = true, transformation(origin = {-2, 0}, extent = {{-10, -10}, {10, 10}}, rotation = 0)));
equation
end mwe;

Simulating this will result in an error message:

[2] 16:56:28 Translation Notification
[mwe: 2:23-2:35]: From here:

[3] 16:56:28 Translation Error
[TestComp: 3:3-3:84]: Non-array modification ‘true‘ for array component ‘fixed‘, possibly due to missing ‘each‘.

Expected Behavior

OMC should detect that the variable testComp.x has type array and needs each.
The resulting model should have the each key word in front of fixed=true:

model mwe
  TestComp testComp(x(each fixed = true, start = {1, 2}))  annotation(
    Placement(visible = true, transformation(origin = {-2, 0}, extent = {{-10, -10}, {10, 10}}, rotation = 0)));
equation
end mwe;

Version and OS

• OpenModelica Version: OpenModelica v1.18.0-dev-232-g187041202b (64-bit)

[AnHeuermann]

@adeas31 I guess ElementProperties.cpp needs to be changed a bit.

[casella]

Duplicate of #5405.

There are several such long-standing issues that really need to be resolved - this one is particularly nasty because it generates invalid code that the new front end does not handle.

@adeas31 do you need changes to the interactive environment to fix this?

[AnHeuermann]

Sorry, I don't see why this is a duplicate of #5405. This issue has nothing to do with the check, the by OMEdit generated code is simply wrong for arrays.

[casella]

I had provided the wrong link, sorry. Now it points where it should.

[adeas31]

If I remember correctly this is because of updateComponent api. The interactive environment need to be fixed first.