__mul__ vs __rmul__ (and others) priority for different classes

[dmitrey]

hi all,
I have created a class MyClass and defined methods like __add__,
__mul__, __pow__, __radd__, __rmul__ etc.
Also, they are defined to work with numbers, Python lists and
numpy.arrays.

Both Python lists and numpy arrays have their own methods __add__,
__mul__, __pow__, __radd__, __rmul__ etc.
If I involve myPythonList * myClassInstance it returns just result of
my __rmul__; but when I invoke nuumpy_arr*myClassInstance it returns
another numpy array with elements [nuumpy_arr[0]*myClassInstance,
nuumpy_arr[1]*myClassInstance, ...].

Can I somehow prevent numpy array of using it __mul__ etc methods?
(and use only my __rmul__, __radd__, etc instead)?

Thank you in advance, D.

 

[Terry Reedy]

hi all,
I have created a class MyClass and defined methods like __add__,
__mul__, __pow__, __radd__, __rmul__ etc.
Also, they are defined to work with numbers, Python lists and
numpy.arrays.

Both Python lists and numpy arrays have their own methods __add__,
__mul__, __pow__, __radd__, __rmul__ etc.
If I involve myPythonList * myClassInstance it returns just result of
my __rmul__; but when I invoke nuumpy_arr*myClassInstance it returns
another numpy array with elements [nuumpy_arr[0]*myClassInstance,
nuumpy_arr[1]*myClassInstance, ...].

Can I somehow prevent numpy array of using it __mul__ etc methods?
(and use only my __rmul__, __radd__, etc instead)?

No. you have to put your class instance first to get priority.
Given a op b, python first calls a.__op__(b) and only if that fails,
which it will for most builtin objects when b is MyClass, b.__rop__(a).
Numpy arrays, however, are more broadminded about what they will work with.

If your operations are not commutative, you will either have to wrap
numpy arrays in a class that disables the special methods or use
explicit function calls.

Terry Jan Reedy