sos: [BUG] Data exchange between Octave and Python3

Hi SoS team!

When I try to pass a 3D matrix (of any type) from Octave to Python3 subkernel, I run into the following problem:

I thought that a smaller 3D matrix would make it, but it still fails:

It works properly, when I reduce it to a 2D matrix by discarding one dimension:

Or when I collapse two dimensions into one:

I would like to note once again that this problem occurs for any data type.

---

• Hoping that passing the 3D matrix in vector form would do the trick, I did the following

However, it gets stuck for a long time and I had to restart the kernel on every attempt.

For now, I will shrink my 3D Octave matrices into 2D and reshape them back to their original form in Python3 . I hope that this issue can help solve the problem. Please let me know if you need further details!

Environment

• Ubuntu 18.04.2 running on Docker
• Python 3.7.3
• Octave 4.2.2
• jupyter 4.4.0
  • notebook 5.7.4
• sos 0.19.8
  • sos-bash 0.18.1
  • sos-javascript 0.18.0
  • sos-julia 0.18.2
  • sos-matlab 0.18.2
  • sos-notebook 0.19.6
  • sos-python 0.18.1
  • sos-r 0.19.1
  • sos-ruby 0.18.1
  • sos-sas 0.18.0