Running MocoInverse solve() with TabOpLowPassFilter crashes Matlab 2019a with R6025 runtime error "Pure virtual function call."

[psbiomech]

Seems to happen with all my data using preprint version 0.1.0, so just adding this in case it's an actual issue and not just me.

[chrisdembia]

Thanks Pras. Could you submit minimal code that reproduces the error?

[chrisdembia]

I’m also curious if this happens in Python or C++.

[psbiomech]

Will cut and paste asap - distracted by a toddler meltdown at the moment. I was just prepping IK/RRA kinematics for MocoInverse. E.g:

mocoinv = MocoInverse();
...
...
kindata = TableProcessor(ikfile);
kindata.append(TabOpLowPassFilter(6.0));
...
...
mocoinv.setKinematics(kindata);
mocoinv.set_kinematics_allow_extra_columns(true);
...
...
solution = mocoinv.solve()

[psbiomech]

Also I'm using MATLAB 2019a. Runs fine if I omit the append call. I set filter frequency to -1 for RRA kinematics.

[chrisdembia]

Pras, thank you for providing more information. I am unable to reproduce this error locally. Can you try upgrading to the 0.2.0 release and see if you still run into this issue?

[psbiomech]

Thanks Chris. Will do asap. Happy new year!

[psbiomech]

Also fyi I don't have this problem running MocoTrack/MocoStudy.

[chrisdembia]

Thanks; that's useful information. I'm still hoping we can reproduce the issue on our end.

[psbiomech]

Unfortunately, I have the same issue in Version 0.2.0 (runtime error R6025) in Matlab 2019a.

With regards to my code above, one thing I didn't mention earlier: the issue only occurs when executing mocoinv.solve(), not when the filter operation is appended to the TableProcessor using TabOpLowPassFilter. If mocoinv.solve() is run without TabOpLowPassFilter, then it runs fine.

Again, no issues with MocoTrack/MocoStudy.

I've retitled the issue to more correctly reflect this.

[chrisdembia]

@psbiomech , using the files you sent locally on my Mac, with Moco 0.2.0, I do not run into a crash (but I do get a NaN error). Perhaps this is a Windows issue. When I'm back in lab next week, I'll try to test this with Nick on his Windows machine.