Consider:

    from collections.abc import *

    class DefinitelyNotAList:
        def __init__(self, backer):
            self.backer = backer

        def __getitem__(self, key):
            return self.backer[key]

        def __len__(self):
            return len(self.backer)

        def __contains__(self, needle):
            return needle in self.backer

        def __reversed__(self):
            return reversed(self.backer)

        def __iter__(self):
            return list.__iter__(self.backer)

        def index(self, *args, **kwargs): return self.backer.index(*args, **kwargs)

        def count(self, *args, **kwargs): return self.backer.count(*args, **kwargs)


    dnal = DefinitelyNotAList([2,4,6,8])

    for abc in [Collection, Reversible, Sized, Iterable, Container, Sequence]:
        print(abc.__name__.ljust(12), isinstance(dnal, abc))


Which prints:

    Collection   True
    Reversible   True
    Sized        True
    Iterable     True
    Container    True
    Sequence     False

I'm not sure whether this is a bug, a docs bug, or just a misunderstanding but my expectation, is that, somehow, I could get isinstance(obj, Sequence) to return *without* explicitly registering it, just as True is returned for the other abcs.

This was tagged as 3.7, but the issue still exists AFAICT in the latest github master: https://github.com/python/cpython/blob/master/Lib/_collections_abc.py

This is an exact duplicate of #35190 - "collections.abc.Sequence cannot be used to test whether a class provides a particular interface (doc issue)"

Edit conflict -- indeed -- self closing.

---

I should point out that this could be fixed with something like the following:

    @classmethod
    def __subclasshook__(cls, C):
        if cls is Sequence:
            return _check_methods(C, "__reversed__", "__iter__",
                                  "__len__", "__contains__", "__getitem__")
        return NotImplemented

But seeing as it's not the only abc that is without a subclass hook, I wanted to raise an issue before I submitted a complete PR implementing subclasshooks for the rest of the abcs that need them.