__array_priority__ regression

[ZacDiggum]

Hi,
I just installed the third v3.3 bugfix release. The problem with adding / multiplying numpy scalars and af arrays somehow came back. See the discussion with @FilipeMaia, issue #69.
Try something like:

a = af.randu(5,5)
b = np.sin(1)*a

b will be a numpy array. Back then, Filipe changed the __array_priority__ attribute.

[ZacDiggum]

I just noticed this only happens with this version which is master, I guess. Devel works fine.

[pavanky]

@ZacDiggum I don't think I ever merged that into master.. That PR was sent to devel.

[ZacDiggum]

:-) Are you going to merge it?

[pavanky]

@ZacDiggum I am going to do it. I have come across a couple of other minor issues as well that need to be resolved.

[pavanky]

I'll close it when the issue is resolved. Makes sure I am keeping track of things :)

[pavanky]

Ok this is now fixed :) sorry about the delay. I hit a bug when testing with unified backend and it ended up being an upstream bug in arrayfire itself :)