| msg289817 - (view) |
Author: Jelle Zijlstra (JelleZijlstra) *  |
Date: 2017-03-18 15:22 |
| pathlib.Path.__new__ takes **kwargs, but doesn't do anything with them (https://github.com/python/cpython/blob/master/Lib/pathlib.py#L979). This doesn't appear to be documented. This feature should presumably be either documented or removed (probably removed unless I'm missing some reason for having it). Brief discussion on a typeshed PR at https://github.com/python/typeshed/pull/991#discussion-diff-105813974R100 |
|
|
| msg289896 - (view) |
Author: Brett Cannon (brett.cannon) * |
Date: 2017-03-20 18:40 |
| Yep, kwargs should be dropped since it isn't used or documented: https://docs.python.org/3/library/pathlib.html#pathlib.PurePath (probably just a hold-over from when it did in some earlier version of the code). |
|
|
| msg289897 - (view) |
Author: Jelle Zijlstra (JelleZijlstra) * |
Date: 2017-03-20 18:47 |
| Thanks, I'll add a PR. This doesn't need to be documented, right? |
|
|
| msg289902 - (view) |
Author: Serhiy Storchaka (serhiy.storchaka) * |
Date: 2017-03-20 19:15 |
| The support of **kwargs in Path.__new__ is needed if you want to implement a subclass of Path with __init__ accepting keyword arguments (and since Path constructor takes variable number of positional arguments, new arguments should be keyword-only). >>> import pathlib >>> class MyPath(pathlib.PosixPath): ... def __init__(self, *args, spam=False): ... self.spam = spam ... >>> p = MyPath('/', spam=True) >>> p MyPath('/') >>> p.spam True Removing **kwargs from Path.__new__ will break the above example. >>> MyPath('/', spam=True) Traceback (most recent call last): File "", line 1, in TypeError: __new__() got an unexpected keyword argument 'spam' |
|
|
| msg289944 - (view) |
Author: Brett Cannon (brett.cannon) * |
Date: 2017-03-21 17:12 |
| Shoot, that's too bad. I guess we should document it then so people are aware that keyword arguments are ignored, else we will break subclasses. There's also an unfortunate difference between PurePath and Path as PurePath doesn't have this quirk. |
|
|
| msg289945 - (view) |
Author: Serhiy Storchaka (serhiy.storchaka) * |
Date: 2017-03-21 17:25 |
| I don't know whether it was the intension of Antoine or just an oversight. I don't know whether it is used in the wild. But we can at least raise a TypeError for concrete classes PosixPath and WindowsPath if ignoring keyword arguments is a problem. Many extension types don't take keyword arguments, but their subclasses accept and ignore keyword arguments. For example: >>> filter(None, [], foo=123) Traceback (most recent call last): File "", line 1, in TypeError: filter() does not take keyword arguments >>> class X(filter): pass ... >>> X(None, [], foo=123) <__main__.X object at 0xb6fdcacc> |
|
|
| msg289946 - (view) |
Author: Antoine Pitrou (pitrou) * |
Date: 2017-03-21 17:32 |
| > The support of **kwargs in Path.__new__ is needed if you want to implement a subclass of Path with __init__ accepting keyword arguments I don't remember exactly, but I think this was the intention indeed. There was originally an openat-using subclass, and IIRC it took additional parameters (such as the directory fd). That got scrapped quite early in the process, so we can remove the **kwargs thing now. |
|
|
| msg289998 - (view) |
Author: Brett Cannon (brett.cannon) * |
Date: 2017-03-22 16:32 |
| Then I vote for Serhiy's idea of simply raising an exception in the concrete subclasses when a keyword argument is given. |
|
|
| msg369520 - (view) |
Author: Rémi Lapeyre (remi.lapeyre) * |
Date: 2020-05-21 11:56 |
| PurePath subclasses cannot support kwargs as __new__() does not accept **kwargs: >>> from pathlib import PurePath >>> class MyPurePath(PurePath): ... def __init__(self, *args, **kargs): pass ... >>> MyPurePath('foo', spam=True) Traceback (most recent call last): File "", line 1, in TypeError: __new__() got an unexpected keyword argument 'spam' The behaviour for this should probably be made the same for both Path and PurePath. |
|
|