SetupExperiment function not longer follow FMI 2.0 Signature

[Jorgelmh]

We've tried to release a new version of PythonFMU, but have encountered a critical error when using FMPy to simulate the PythonFMU generated FMU:

I've looked at the FMI standard 2.0 specification and it seems the function signature is no longer following the specification. SetupExperiment should be able to accomodate these parameters:

This used to be okay in previous versions, but it seems that new work on the native modules have broken this functionality.

I suggest that we also add a simple test that steps an FMU generated with PythonFMU using FMPy, since this is a popular tool choice for simulating FMUs.

[markaren]

Hi, the issue is that the code overriding pythonfmus setupExperiment must be updated with the new signature. PythonFMU itself has no issues wrt the FMI spec.

[markaren]

However, a backwards compatible solution would be nice to have.

[eisDNV]

Could you please explain what you mean with 'the code overriding pythonfmus setupExperiment must be updated'? I thought that the function 'Fmi2Slave.setup_experiment( start_time)' only needs to be defined if one wants to achieve something 'non-default', while the current update requires that all existing models based on PythonFMU must be updated before they can be re-compiled. I did not find any instructions that this must be done, nor is it easy to understand why this change was necessary and what we gain with it? I would really like if that feature could be reverted.

[markaren]

The change came in https://github.com/NTNU-IHB/PythonFMU/pull/246/files and should only affect code that overrides setup_experiment. But it is a breaking change. If user code uses setup_experiment, then it must be updated to take 3 parameters not 1.