In-place methods in subclasses

Question

I am trying to subclass list in order to create an in-place method that updates self with the item-wise sum of itself and a new list. Example:

This is how I tried to do it:

class RecordBook(list):
    def accumulate(self, new_values):
        self = RecordBook([x+y for x, y in zip(self, new_values)])

With the though being that rebinding back to self will modify it in-place. It does not work however.

>> d = RecordBook([1, 2, 3, 4, 5, 6])
>> d
[1, 2, 3, 4, 5, 6]
>> d.accumulate([1]*6)
>> d
[1, 2, 3, 4, 5, 6]  # should have been [2, 3, 4, 5, 6, 7]

Why is that failing the task and how would one go about this?

You're not *modifying* the instance, you're trying to *replace* the instance with a completely new instance. Don't do that. You want something like `self[:] = [...]` to modify an existing list. — deceze, Oct 13 '20 at 06:58
@deceze You are totally right but I still don't get why it did not work — Ma0, Oct 13 '20 at 07:00
You probably want to subclass [`collections.UserList`](https://docs.python.org/3/library/collections.html#collections.UserList) i/o `list` — Reblochon Masque, Oct 13 '20 at 07:01
Because `self` is just a local variable, and assigning something else to it doesn't replace the existing object. `a = 'foo'; b = a; b = 42` — Did that replace or modify `a`? Nope. — deceze, Oct 13 '20 at 07:03
"but I still don't get why it did not work" Because nothing in Python ever works that way. If you do `a = []` and `b = a` and then `b = 3`, `a` does not become `3`. — Karl Knechtel, Oct 13 '20 at 07:16
@KarlKnechtel If these two examples are equivalent, my understanding of `self` was flawed. Thanks! — Ma0, Oct 13 '20 at 07:49
`self` is just another parameter. The magic is done [elsewhere](https://docs.python.org/3/howto/descriptor.html). — Karl Knechtel, Oct 13 '20 at 10:50

Reblochon Masque · Answer 1 · 2020-10-13T07:12:23.330

Here is one approach subclassing collections.UserList, and modifying the underlying self.data of the subclass.

As explained by @Deceze in the comments:

self is just a local variable, and assigning something else to it doesn't replace the existing object

UserList exposes the self.data attribute that can be modified "in-place"

from collections import UserList


class Accumulator(UserList):
    def accumulate(self, seq):
        for idx, elt in enumerate(seq):
            self.data[idx] += elt


if __name__ == '__main__':
    d = Accumulator([1, 2, 3, 4, 5, 6])
    d.accumulate([1]*6)
    print(d)

output:

[2, 3, 4, 5, 6, 7]

In-place methods in subclasses

1 Answers1

output: