Allow FMUs to require start values about fmi-cross-check

[KarlWernersson]

The default start values given in the xml has to be included in the compiled or the source code fmu.
It would be much easier to require FMU's to be exported with the default start values. For me this is obvious but maybe we need to spell it out?

[KarlWernersson]

As there is no current way of specifying different start values of the FMU in the cross check, an FMU that requires a different set than the default is not compliant with the cross check rules

[t-sommer]

The point is to test if different start values are set correctly by the importing tool. If an FMU has different start values compiled in than what is in the XML it's simply illegal and should be excluded from the cross-check.

[KarlWernersson]

Would people export such fmu's? Potentially make it harder for them to get passed by vendors not handling it correctly?
How much extra work would it be for importers to handle this in the import?
I see it more of interest as a reference fmu test rather than a compliance test.

[chrbertsch]

This would indeed be an ideal case for "ReferenceFMUs" not exported by a commercial tool, but with the expectation that every tool can simulate them correctly (or explain why start value cannot be set in this tool). But in order to include this in the XC, we have to extend the XC infrastructure to support this.

[t-sommer]

That is exactly the intention of the Feedthrough Test-FMU.

[t-sommer]

Discussion at Design Meeting Regensburg: Good idea.
Should be limited it to fixed and tunable parameters.