Run doesn't create new variables in the octave environment about oct2py HOT 3 OPEN

Hi @WiseroOrb, thanks for the kind words.

Yes, this is an unfortunate limitation of the way we're evaluating the script. If you need a result you'd have to use a function. Something like:

test.m

function a = test()
   a = [1,2,3];
end

test.py

from oct2py import octave as oc
a = oc.test()
print(a)

Which gives: [[1. 2. 3.]]

from oct2py.

Hi @blink1073 ,
So I'm in the process of converting the script in the way you suggested, however I've encountered a major roadblock: I need to return an object from the statistical package, of class gmdistribution. Oct2py trows some warning when I try to return it, and later fails to use it in other functions.
I think the origin of the problem is because this class doesn't implement saveobj.

Right now I'm planning to do use eval(distribution = computeDistribution(args)) and keeping the distribution object in the octave workspace.

This pollutes the way I call the functions, because from now on I need to push/pull every argument and use the eval statement every time I need to use the distribution object.

I thought of treating it as a global variable and not as a funtion paramenter, but this requires writing wrappers for each affected function.

So, python side:

oc.eval('global distribution') #after computeDistribution
oc.fooModified(arg)

Wrapper function

funtion out = fooModified(arg)
    global classifier
    out = foo(arg, classifier)
end

Do you have any better solution?

from oct2py.

The only other option I can see if there is if you can serialize the gmdistribution data yourself by extracting properties into a cell object.

from oct2py.