http://docs.python.org/3/library/functions.html#issubclass

"classinfo may be a tuple of class objects."

I suggest make it clear like isinstance, use something like this:

classinfo may be a tuple of class objects and such tuples.

While this is documented for isinstance, I'm not sure it should be advertised too much, as it seems to me an implementation detail and doesn't seem too useful in practice.

This is a side-effect of the fact that
isinstance(x, (A, B, ...))
is equivalent to
isinstance(x, A) or isinstance(x, B) or ...
and therefore
isinstance(x, (A, (B, C)))
is equivalent to
isinstance(x, A) or isinstance(x, (B, C))
which in turn is equivalent to
isinstance(x, A) or (isinstance(x, B) or isinstance(x, C))

While this behavior seems intentional 0, it doesn't seem to be tested 1. FTR this is supported by PyPy too.

Whether the recursive nature of the check should be documented or not, it seems inconsistent that it's documented for isinstance but not for issubclass.

@Irit, you may be interested in bpo-44135, there's an open PR there and it seems to fix what this issue describes.

@ken - not quite. This issue is about isinstance containing "(or recursively, other such tuples)" which is not there for issubclass.

This was made consistent by #92087 - if the recursive wording shouldn't be there for both funcs, then I think a new issue could be opened to discuss that instead of here.