I was looking to see if there are equivalents to Java’s private and protected members, and it looks like Python’s answer to that is just throw one or two underscores in front of things to do that. And it doesn’t really do anything, more of just a naming convention. To me that feels like a basic OO structure that is shoehorned into Python.
A single underscore is just a naming convention, but double underscores triggers automatic name-mangling of the variable in question:
$ cat test.py
classfoo:def__init__(self, x):
self.__x = x
f = foo(1)
f.__x
$ python3 test.py
Traceback (most recent call last):
File"/mnt/d/test.py", line 6, in <module>
f.__x
AttributeError: 'foo' object has no attribute '__x'
However, much like private/protected variables in java, this is pretty trivial to circumvent if you want.
But I don’t believe that you can argue that access modifiers are required for OO not to be shoehorned into a language, not when influential OO languages like Smalltalk didn’t have this feature either. Java just happens to be closer to C++, where public/private/protected is much more rigidly enforced than either Java or Python
Yeah, some weird accusations. Python has had classes since its inception (1.0).
Also the image in the post makes no sense. It shows multiple (Spidey) instances all pointing to each other which is not how self works.
self is just a parameter that may contain different instances depending how it was called. This is also true for any other parameters in any function, each time a function is called it may have a different instance.
In what way does OOP feel shoehorned in with Python? I ask since that is not my own impression of the language.
Would you also be willing to share what language(s) you feel do(es) OOP without it being shoehorned in?
I was looking to see if there are equivalents to Java’s private and protected members, and it looks like Python’s answer to that is just throw one or two underscores in front of things to do that. And it doesn’t really do anything, more of just a naming convention. To me that feels like a basic OO structure that is shoehorned into Python.
A single underscore is just a naming convention, but double underscores triggers automatic name-mangling of the variable in question:
$ cat test.py class foo: def __init__(self, x): self.__x = x f = foo(1) f.__x $ python3 test.py Traceback (most recent call last): File "/mnt/d/test.py", line 6, in <module> f.__x AttributeError: 'foo' object has no attribute '__x'However, much like private/protected variables in java, this is pretty trivial to circumvent if you want.
But I don’t believe that you can argue that access modifiers are required for OO not to be shoehorned into a language, not when influential OO languages like Smalltalk didn’t have this feature either. Java just happens to be closer to C++, where public/private/protected is much more rigidly enforced than either Java or Python
Yeah, some weird accusations. Python has had classes since its inception (1.0).
Also the image in the post makes no sense. It shows multiple (Spidey) instances all pointing to each other which is not how self works. self is just a parameter that may contain different instances depending how it was called. This is also true for any other parameters in any function, each time a function is called it may have a different instance.