[Bug] Memory issue using Record with Array

[Vincouux]

Version

OS: Ubuntu 22.04 Python: 3.10.2 Pulsar: pulsar-client==3.2.0

Minimal reproduce step

from pulsar.schema import Record, Array, Integer

class A(Record):
    a = Array(Integer(), required=False, default=[])

# Create a first instance of object A
a = A()
a.a # Correctly show empty list
a.a.append(1) # Adding an integer
a.a # Correctly showing list with single element 1 inside

# Instanciating a second object A
b = A()
b.a # Incorrectly show list with single element 1

What did you expect to see?

I expected the A class not to re-use the previous reference to the list.

What did you see instead?

As a developer, instantiating a fresh object in Python generally means it's built from scratch. Therefore, no reference to previously built objects should be used.

[gromsterus]

1. list is a mutable data structure in Python. In your example, new instances of class A reference and modify the same list object in memory. You can read more about this here.

2. The Record class has quite simple logic for working with default. As a solution, you could add a default_factory implementation, like here. Or, do not use default and instantiate objects of the Record class by passing arguments manually.

[Vincouux]

I decided to remove default and making every mutable data structures required as a solution ! Thanks for the answer.