StateSelect Attribute with unfixed parameters wrong evaluated without warning

[Jens Frenkel]

Consider the following model. The state select attribute is set via unfixed parameters. The FrontEnd evaluates the unfixed parameters but report no warning about that. The result is totaly different to the exprected from the modelers view for y and z.

model StateSelectCheck
  parameter Boolean preferredStates=true;
  parameter Boolean preferredStatesUnfixed(fixed=false);
  parameter Boolean preferredStatesUnfixedStart(fixed=false,start=true);
  Real x(stateSelect = if preferredStates then StateSelect.prefer else StateSelect.avoid);
  Real y(stateSelect = if preferredStatesUnfixed then StateSelect.prefer else StateSelect.avoid);
  Real z(stateSelect = if preferredStatesUnfixedStart then StateSelect.prefer else StateSelect.avoid);
initial equation
  preferredStatesUnfixed = true;
  preferredStatesUnfixedStart = false;
equation
  der(z) = time;
  0 = x^2 + y^2 + z^2;
  y = x + z;
end StateSelectCheck;

Result after the FrontEnd:

class StateSelectCheck
  parameter Boolean preferredStates = true;
  parameter Boolean preferredStatesUnfixed(fixed = false);
  parameter Boolean preferredStatesUnfixedStart(start = true, fixed = false);
  Real x(StateSelect = StateSelect.prefer);
  Real y;
  Real z(StateSelect = StateSelect.prefer);
initial equation
  preferredStatesUnfixed = true;
  preferredStatesUnfixedStart = false;
equation
  der(z) = time;
  0.0 = x ^ 2.0 + y ^ 2.0 + z ^ 2.0;
  y = x + z;
end StateSelectCheck;

[Martin Sjölund]

Note that it is valid for a tool to reject non-fixed structural parameters. That is, if the parameter cannot be evaluated like a constant, we could reject the model.

[Martin Sjölund]

Postponed until 1.9.1

[Martin Sjölund]

This ticket was not closed for 1.9.1, which has now been released. It was batch modified for milestone 1.9.2 (but maybe an empty milestone was more appropriate; feel free to change it).

[Martin Sjölund]

Milestone changed to 1.9.3 since 1.9.2 was released.

[Martin Sjölund]

Moved to new milestone 1.9.4

[Martin Sjölund]

Milestone pushed to 1.9.5

[Martin Sjölund]

Milestone renamed

[Martin Sjölund]

Ticket retargeted after milestone closed

[Martin Sjölund]

Milestone moved to 1.12.0 due to 1.11.0 already being released.

[Lennart Ochel]

Yes, that model should be rejected. But in fact, the state select attribute of y and z is ignored, i.e. removed, without any feedback.

[Francesco Casella]

As of v1.13.0-dev-155-g68350e9, using the new front end, the translation fails with

[1] 01:35:33 Translation Error
Internal error Instantiation of M failed with no error message.

[2] 01:35:34 Scripting Error
[C:/dev/OpenModelica64bit/OMCompiler/Compiler/NFFrontEnd/NFFlatten.mo:1314:9-1314:72:writable]Modelica Assert: NFFlatten.makeStateSelectAttribute got non StateSelect value! 

I guess this is the expected behaviour, even though the error message is not that clear. @perost can you please check?

[Francesco Casella]

The error message is now

[1] 19:11:38 Translation Error
[C:/dev/OpenModelica64bit/OMCompiler/Compiler/NFFrontEnd/NFCeval.mo: 1096:9-1097:96]:
Internal error NFCeval.evalIfExp:
unimplemented case for if preferredStatesUnfixed then StateSelect.prefer else StateSelect.avoid

I guess a clearer error message is still needed.

[Francesco Casella]

Still valid as of v1.13.0-dev-798-g1c8bb86de

[Per Östlund]

Replying to casella:

Still valid as of v1.13.0-dev-798-g1c8bb86de

The NF actually detected the error, but the logic for when to print an error message was wrong. With the fix in cc477d0 the model now fails with:

[test.mo:3:3-3:56:writable] Notification: From here:
[test.mo:6:10-6:96:writable] Error: Constant preferredStatesUnfixed is used without having been given a value.

Good enough?

[Francesco Casella]

Excellent!