Prepare for CI: Magnetic.QuasiStatic.FundamentalWave by AHaumer · Pull Request #4787 · modelica/ModelicaStandardLibrary

AHaumer · 2026-05-03T09:17:11Z

same procedure as #4775, except Modelica.Magnetic.QuasiStatic.FundamentalWave.Examples.BasicMachines.InductionMachines.IMC_withLosses where IMHO a longer interval (factor 100) is sufficient to reduce the size of the reference results.

…tic.FundamentalWave

christiankral · 2026-05-03T20:08:00Z

@AHaumer Some "new" models have been added to ModelicaTest. Why can't we just extend these models from the MSL and add a proper annotation? They look quite identical to me.

christiankral · 2026-05-04T07:10:33Z

Extending from the Modelica Standard Library make the code very maintainable. One more aspect here:

The annotations in ModelicaTest now show simulation start and stop time and indicate the time window for comparison. This is OK.
The same annotation also shows the documentation of original model stored in the MSL. This causes some obvious conflicts. Example:

 extends Modelica.Magnetic.QuasiStatic.FundamentalWave.Examples.BasicMachines.SynchronousMachines.SMPM_CurrentSource;
  annotation (
    experiment(StopTime=0.20, Interval=1E-4, Tolerance=1E-6),
    TestCase(shouldPass = true,
      __ModelicaAssociation(Comparison(TimeWindows={TimeSlot(0.00, 0.20)}))),
    Documentation(info="<html>
<p>
This example compares a time transient and a quasi-static model of a permanent magnet synchronous machine. The machines are fed by a current source. The current components are oriented at the magnetic field orientation and transformed to the stator fixed reference frame. This way the machines are operated at constant torque. The machines start to accelerate from standstill.</p>

<p>
Simulate for 2 seconds and plot (versus time):
</p>

The stop time is set to 0.20 seconds and the documentation states "Simulate for 2 seconds".

I wonder if it makes more sense to just state something like:

    Documentation(info="<html>
<p>
The original documentation is available at the model from which this one is extended.
</p>

Reasons:

Who is maintaining the documentation of the ModelicaTest models extended from the MSL? I guess no one. So whenever we update the documentation of the MSL model, it is entirely forgotten about the documentation of the ModelicaTest models.
However, we could add a hyperlink to the original model of the MSL. But this is not required from my point of view, as the extends clearly shows which model we refer to.

henrikt-ma · 2026-05-04T07:21:40Z

The annotations in ModelicaTest now show simulation start and stop time and indicate the time window for comparison. This is OK.

I don't see this being OK regarding the StartTime, as this may have all sorts of unwanted effects compared to just trimming the first part of the main model's simulation result.

To make time windowed comparison beyond just taking the initial part of the result of a chaotic system, we need this TestCase annotation that we have talked about, where the time intervals have both beginning and end.

AHaumer · 2026-05-04T08:28:46Z

@christiankral I adreesed your concerns. Sorry for not using extends, my fault.
@henrikt-ma I don't understand your comment.
All the "extended" examples in ModelicaTest use the same default StartTime=0 as the original models.

henrikt-ma · 2026-05-04T09:55:50Z

@henrikt-ma I don't understand your comment. All the "extended" examples in ModelicaTest use the same default StartTime=0 as the original models.

I was commenting on the following (emphasis is mine):

The annotations in ModelicaTest now show simulation start and stop time and indicate the time window for comparison. This is OK.

That is, if you remove the start and, and maybe stop calling it a time window when the window must be open-eded to the left, then it would seem OK to me as well.

I would rather call what you are doing something like a short-time comparison model to not cause confusion with a proper time-windowed comparison which we seem to need sooner or later anyway.

HansOlsson · 2026-05-04T11:13:15Z

I wonder if it makes more sense to just state something like:
 Documentation(info="<html>

The original documentation is available at the model from which this one is extended.

Reasons:

Who is maintaining the documentation of the ModelicaTest models extended from the MSL? I guess no one. So whenever we update the documentation of the MSL model, it is entirely forgotten about the documentation of the ModelicaTest models.

However, we could add a hyperlink to the original model of the MSL. But this is not required from my point of view, as the extends clearly shows which model we refer to.

I agree that we shouldn't duplicate the documentation, but to me it seems simplest to as default not have any documentation for these models.

Tools can already show that it extends from another model, and in some way make that accessible.

In Dymola that would currently give something like:

Information

Extends from Modelica.Magnetic.QuasiStatic.FundamentalWave.Examples.BasicMachines.SynchronousMachines.SMPM_CurrentSource
(Test example: PermanentMagnetSynchronousMachine fed by current source).

(But with an internal link instead.)

Thus, I don't see that spending time on writing "The original documentation is available at the model from which this one is extended." really helps.

I could even imagine that tools could be more helpful if the documentation of the derived class is empty, so it might be that the text is counter-productive.

HansOlsson · 2026-05-04T11:15:56Z

+    Documentation(
+        info="<html>
+<p>
+The original documentation is available at the model from which this one is extended.
+</p>
+</html>"),
