Use of undeclared identifier and undefined reference to *_array_get function

[federico.terraneo@…]

The following model (Top in the Bug package) fails compilation.
For some reason, if z is 1, omc produces C code including calls to a nonexistent function Bug_Component_array_get. The calls to this function also reference a parameter components which does not exist in the caller's context.

Setting z to anything greater than 1 produces working code, where the Bug_Component_array_get is not declared not used anywhere.

[federico.terraneo@…]

Code causing the compilation error

[Francesco Casella]

@mahge930, could you have a quick look?

[Mahder Alemseged Gebremedhin]

in model Top the value of col is provided by the declaration

model Top
  parameter Integer z = 1;
  Component components[x](each cols = z);
  ...
end Top;

The array T in model Component is declared as

model Component
  ....
  Modelica.SIunits.Temperature T[rows, cols]
  ...
equation
  ...
  c * der(T[1, 1]) = gx * (T[1, 2] - T[1, 1])
                        + gy * (T[2, 1] - T[1, 1])
                        + gf * (inletFluidTemperature - T[1, 1])
                        + bottom.port[1].Q_flow;
  

As you can see, the second equation in model Component tries to access T[1,2] regardless of the value of cols.

This is where the problem is. If z is 1 then cols is set to 1 which means T is a [4,1] array. Trying to access [1,2] should normally cause out of range access error.

The way our simulation code operates right now, if it fails to find a variable in the list of simcode variables, in this case components[1].T[1,2] for example, it will assume it is a temporary variable and tries to generate function like code for it. You don't have to worry about the details normally. Until you run in to this kinds of odd errors.

I understand it would be nice to have better error messages in this case. However this is not something we can fix right now. At least not as a quick fix. For what it is worth I think this error is actually a bit better than having a silent out of bounds access.

[federico.terraneo@…]

Thanks for looking into that, mahge930.
The lack of a compile time or run-time error message for index out of bounds made me think it was a compiler error.
I guess OpenModelica's trac is indexed by search engines, so if other users run into the same problem they can find this ticket and know they have to carefully check their models.

From my side it's fine to close the ticket if you want.

[anonymous]

I noticed another issue, though: if one simulates the Component model directly, with cols=1 then omc does not produce any error about the index out of bounds.

I'm attaching a model showing the issue. It's just the same model as before without Top, and with cols in Component set to 1.

In this case, the access to T[1,2] is out of bounds and should produce some kind of error.

[federico.terraneo@…]

Undetected index out of bounds

[Francesco Casella]

Retargeted to 1.17.0 after 1.16.0 release

[Francesco Casella]

Retargeted to 1.18.0 because of 1.17.0 timed release.

[Francesco Casella]

Ticket retargeted after milestone closed