>>> x = io.StringIO(1)
>>> x.read()
'1'

I believe you're referring to StringIO, if so, it changes the parameter
received to a string:

"""
class StringIO:
  def __init__(self, buf = ''):
        # Force self.buf to be a string or unicode
        if not isinstance(buf, basestring):
            buf = str(buf)
"""

oops, stupid me, this a 3.0 issue..., well seems the str() conversion is
done as well on the 3.0 io module:

"""
class StringIO(TextIOWrapper):
   def __init__(self, initial_value="", encoding="utf-8",
                 errors="strict", newline="\n"):
        super(StringIO, self).__init__(BytesIO(),
                                       encoding=encoding,
                                       errors=errors,
                                       newline=newline)
        if initial_value:
            if not isinstance(initial_value, str):
                initial_value = str(initial_value)
"""

In Python 2.x StringIO.StringIO calls str() on its arguments:

>>> StringIO.StringIO(1).read()
'1'
>>> StringIO.StringIO(object).read()
"<type 'object'>"
>>> str(object)
"<type 'object'>"

io.StringIO has the same behavior:
>>> io.StringIO(1).read()
'1'
>>> io.StringIO(object).read()
"<type 'object'>"

Georg, what's the bug here?

According to the docs: "When a StringIO object is created, it can be
initialized to an existing string by passing the string to the constructor."
So, it seems that that the docs are being too strict, or the code is
being too open. The conversion to str in the constructor is explicit, so
I'd favor leaving it be.

Okay, then I see this as a intended feature.