I think I can learn a lot from any answer to the two questions in this post.

OK, I’ve got a large object. That object has attributes and does things.

Those attributes also do things. MOST of what they do is sit there like a list, int, or np.array and wait to tell something else what to do. BUT SOMETIMES, they’ve got to do something on their own.

So, I can’t do, for instance

if method.attribute == something:

I’ve got to instead do

If method.attribute.value == something:

I can get around this with lists or ints by making a class with a “default” value and additional methods:

class Int(int):pass

But I can’t do that with arrays, so I sometimes end up with ugly stuff like

if method.attribute.value(key).value == something:

So, two questions:

Have I gone completely wrong?

If not, what is the canonical way to guard against initializing a new instance with

NewInstance = ClassName(OldInstance, *args)

when I meant

NewInstance = ClassName(OldInstance.value, *args)

Right now, I’m doing something like

class ClassName(object):
def __init__(self, value, *args):
        self IsClassNameInstance = True
        if hasattr(value, ‘IsClassNameInstance’):
            raise # some error

Thank you.

5 Years
Discussion Span
Last Post by shayallenhill

Could you give short example with one type and some values to show what you are really doing, I do not understand. (So otherwise I would say you are completely wrong, but I am not really OO expert)


Thanks for the reply. Here's an example:

A Mesh object storing and manipulating Points, Edges, and Faces in a half-edge data structure.

The Mesh object stores mesh data as a dict of Points, a dict of Edges, and a dict of Faces. Let's just talk about the dict of Points (call it PDict).
Each Point in PDict is a numpy.array describing a location in 3D space. One of the methods which acts upon these Points must examine all Mesh data then select and combine some of the Points, so I create a method, combine, of Mesh which examines Edges and Faces and selects and combines certain Points in PDict. The selection of Points is described in vombine. The combination of Points is simple:


no method required.

So, the furthest I need to reach into my object is


Textbook OOP in my limited understanding.

BUT, say I want to assign some characteristics to each Point. Now, each Point has to go from /being/ a np.array to /having/ a np.array. Now, I have to reach a little deeper to get to the location of a Point


And the combination of points is no longer simple. I've got to combine the points /and/ their characteristics in a combine method within my point class.

And the combination of each characteristic is not simple, either. So each characteristic /has/ an int or /has/ a list or /has/ an array along with its own combine method.

Now, if I must reach a Point characteristic from Mesh, I have to


And it just gets uglier from there, which is why I suspect I'm doing something wrong. This should be an ideal OOP task, but the procedural version (tests make adding new Point characteristics easy) comes out simpler.

This topic has been dead for over six months. Start a new discussion instead.
Have something to contribute to this discussion? Please be thoughtful, detailed and courteous, and be sure to adhere to our posting rules.