Library error when importing scipy.optimize in embedpy

[mcg1969]

cc: @ilanschnell

I'm working on the embedPy recipe right now, and it was my desire to have conda build run the tests. I've run into the following problem. I have an environment with the current build of kdb and embedpy. When running the curve_fit test I'm getting a library loading error; see below. I'm continuing to investigate, but if the embedPy authors have seen this before, let me know.

From the command line:

conda create -n kx -c kx kdb embedpy scipy
conda activate kx
q

From the q prompt:

\l p.q
cf:.p.import[`scipy.optimize;`:curve_fit;<]

The error:

'import: dlopen(/Users/mgrant/miniconda3/envs/kx/lib/python3.6/site-packages/scipy/optimize/minpack2.cpython-36m-darwin.so, 2): Symbol not found: _main
  Referenced from: /Users/mgrant/miniconda3/envs/kx/lib/python3.6/site-packages/scipy/optimize/minpack2.cpython-36m-darwin.so
  Expected in: flat namespace
 in /Users/mgrant/miniconda3/envs/kx/lib/python3.6/site-packages/scipy/optimize/minpack2.cpython-36m-darwin.so
  [1]  /Users/mgrant/miniconda3/envs/kx/q/p.q:36: .p.wfunc:{[f;x]r:wrap f x 0;$[count x:1_x;.[;x];]r}

[mcg1969]

Incidentally, the curve_fit tests are the only ones failing for me.

[mcg1969]

Oh, and I should add that importing scipy.optimize.curve_fit within Python itself works just fine.