How to use the fitsio.util.is_object function in fitsio

The column in which to write.  Can be the name or number (0 offset)
        column: ndarray
            Numerical python array to write.  This must be an object array.
        firstrow: integer, optional
            At which row you should begin writing.  Be sure you know what you
            are doing!  For appending see the append() method.  Default 0.
        """

        if not keys:
            import warnings
            warnings.warn(
                "The keyword arguments '%s' are being ignored! This warning "
                "will be an error in a future version of `fitsio`!",
                DeprecationWarning)

        if not is_object(data):
            raise ValueError("Only object fields can be written to "
                             "variable-length arrays")
        colnum = self._extract_colnum(column)

        self._FITS.write_var_column(self._ext+1, colnum+1, data,
                                    firstrow=firstrow+1)
        self._update_info()

elif names is not None:
                    columns_all = names
                else:
                    raise ValueError(
                        "you must send `columns` or `names` "
                        "with a list of arrays")
            else:
                columns_all = list(data.keys())
                data_list = [data[n] for n in columns_all]

            colnums_all = [self._extract_colnum(c) for c in columns_all]
            names = [self.get_colname(c) for c in colnums_all]

            isobj = numpy.zeros(len(data_list), dtype=numpy.bool)
            for i in xrange(len(data_list)):
                isobj[i] = is_object(data_list[i])

        else:
            if data.dtype.fields is None:
                raise ValueError("You are writing to a table, so I expected "
                                 "an array with fields as input. If you want "
                                 "to write a simple array, you should use "
                                 "write_column to write to a single column, "
                                 "or instead write to an image hdu")

            if data.shape is ():
                raise ValueError("cannot write data with shape ()")

            isrec = True
            names = data.dtype.names
            # only write object types (variable-length columns) after
            # writing the main table