Updated test FMUs with OpenModelica v1.18.0-dev.192

[AnHeuermann]

No description provided.

[lochel]

What version of OpenModelica did you use? 1.17.0 is already too old. It should be OMCompiler v1.18.0-dev.171+gc263934f53 or never.

[lochel]

And did you do the submodule updates on purpose or by mistake?

[AnHeuermann]

What version of OpenModelica did you use? 1.17.0 is already too old. It should be OMCompiler v1.18.0-dev.171+gc263934f53 or never.

Then I'll use todays master.

And did you do the submodule updates on purpose or by mistake?

Dang it...

[lochel]

Looks good with all the warnings gone 😃

[AnHeuermann]

There is one change in https://github.com/OpenModelica/OMSimulator/pull/999/files#diff-3e28691a2c7245ed9450a4ec5db2a2633eda524f6c05791acfdacdb50e4e3709L138-R123.
The values of limiter.u and limiter.y changed by 0.1.

[lochel]

Which test case is it? Your link doesn't work for me.

[AnHeuermann]

testsuite/OMSimulator/PI_Controller.lua

[lochel]

I guess you refer to PI_Controller.lua. This is fine. The new results are better due to proper initialization 👍

[AnHeuermann]

AircraftVehicleDemonstrator is failing, but I didn't change that FMUs.
Any ideas @lochel?

[lochel]

Those FMUs don't run on 32bit Windows. However, I have no clue why it also failed on 64bit Windows....

[arun3688]

@AnHeuermann I tested your PR in my computer and the AircraftVehicleDemonstrator worked for me

[AnHeuermann]

Now it worked. Should we merge it?

[lochel]

@adrpo, @sjoelund It seems that the behaviour is different on different Windows machines in our jenkins cluster.

[sjoelund]

@adrpo, @sjoelund It seems that the behaviour is different on different Windows machines in our jenkins cluster.

So it fails on the real Windows and succeeded on the virtual machine?

[AnHeuermann]

Do we have some time limits for the test? The AircraftVehicleDemonstrator test is slow:
On my Windows with OMDev and OMSimulator build as standalone I need around 215 seconds with both clang and gcc.
But both are working.

[adrpo]

Testing using rtest on windows machines is really very unreliable. rtest hangs sometimes, other issues pop up, etc.

[adrpo]

For example, on windows OMSimulator hanged 20 hours already. In general is perl that hangs running rtest and you need to kill it manually.
https://test.openmodelica.org/jenkins/blue/organizations/jenkins/OMSimulator/detail/PR-999/2/pipeline

[arun3688]

@AnHeuermann I did not measure the time, but in my computer also it took close to 5 minutes to finish

[lochel]

I think the problem here is more related to numeric. The example finished in both cases, but with slightly different numeric results. So it might be a different boost version or any other lib?! Not really sure though.