Model simulation breaks since commit 7dfbb01

[palm86@…]

OM properly simulates my model up until commit c3dd385, but breaks on 7dfbb01. I have tried to pull out as many parts of my model as possible to arrive at a minimal failing model. Attached is my attempt. I realize that it is still rather big.

I've noticed that if I substitute a certain parameter with its value (which happens to be 0.0), the model simulates again. Mathematically the models are identical. Attached is the original model (TransformationUnsorted.mo) and also the model in which the parameter has been substituted with its value (TransformationSubstituted.mo).

Moreover, if I sort the variable and parameter declarations, the model results are different. Attached is also the TransformationSorted.mo file.

I apologize for the large model, I tried to reduce it even further, but the bug started disappearing after making even seemingly simple replacements, for instance, consider equations of the form:

a + b = 0

Replacing a with -b, would sometimes "fix" the model again.

[palm86@…]

This is the same issue I reported here: ​https://www.openmodelica.org/forum/default-topic/1766-model-failing-since-omcompiler-commit-56d186b60d7904ed253b34995e6ff454

[Patrick Täuber]

This issue seems to be fixed by now.