__rmul__ did not act as expected

[songjhaha]

Here is a small demo to reproduce the problem:

using PythonCall
struct MyArray
    x::Array
end
function Base.:*(a::MyArray, b::AbstractArray)
    println("calling *(MyArray, AbstractArray)")
    return a.x * b
end
function Base.:*(a::AbstractArray, b::MyArray)
    println("calling *(AbstractArray, MyArray)")
    return a * b.x
end
a = [1 2; 3 4]
b = MyArray([4 3; 2 1])
a * b
# [8 5; 20 13]

pya = pyjlraw(a) # try not to use pya.__mul__()
pyb = pyconvert(Py, b)

# it's calling pyb.__rmul__(pya)
pya * pyb
# calling *(MyArray, AbstractArray)
# Python ArrayValue: <jl [13 20; 5 8]>

looks like pyb.__rmul__(pya) is calling *(MyArray, AbstractArray), so gets wrong result

[cjdoris]

Thanks - all the reversed operands were not actually reversing their operands! This is fixed on the main branch now.