Original ticket http://projects.scipy.org/numpy/ticket/1739 on 2011-02-10 by @pv, assigned to unknown.

Interestingly,

>>> np.issubdtype(int, np.bool_)
False
>>> np.issubdtype(str, bool)
True

The issubdtype routine apparently does not work correctly for this case (and never has). What it does is

    if issubclass_(arg2, generic):
        return issubclass(dtype(arg1).type, arg2)
    mro = dtype(arg2).type.mro()
    if len(mro) > 1:
        val = mro[1]
    else:
        val = mro[0]
    return issubclass(dtype(arg1).type, val)

Using mro[1] makes e.g. the Python int type behave as the more general signedinteger.

For bool, it's directly generic, which results into strangeness.

It might be better to hardcode the special handling of Python types, than just picking mro[1]...