returnArray cause segmentation fault when using cdfMod "scipy"

[try2code]

returnArray cause segmentation fault when using cdfMod "scipy"

[martinclaus]

This is related to scipy/scipy#3630
I already fixed it in my fork, push request is coming up.

[Try2Code]

Thanks for your work! The changes look great

[Try2Code]

I took over your changes without further testing. Meanwhile I merged some python3 related changes in, but I could not get rid off a segvault when using python2 and scipy in the test_phc test from the test suite.

test_phc is not an official test, but I'd like to fix it, too. Could you give me a hint, that goes wrong?

Thanks in advance!

[martinclaus]

I tried to reproduce the segvault with 739db2c, but I can't. I crafted a test dataset using following ferret script

def ax/x=-180E:180E:5 lon
def ax/y=90S:90N:5 lat
def ax/z=0:1000:200 depth
def gr/x=lon/y=lat/z=depth gout
set gr gout
let TempO = IF i/2 EQ int(i/2) THEN x+y+z
let SO = IF i/2 eq int(i/2) THEN x-y+z
cancel mode upcase_output
save/file=test_phc.nc/clobber TempO, SO

With this dataset, the test runs perfectly. Can you find out, which line in test_cdo.py is actually producing the segmentation fault?

[Try2Code]

changes available in the new release 1.2.4

thanks for your help again!