DOC: numpy.lib.npyio._read refers to read_from_filename, which does not exists.

[Carreau]

Issue with current documentation:

Those two examples are present:

    Examples
    --------
    First we create a file for the example.

    >>> s1 = '1.0,2.0,3.0\n4.0,5.0,6.0\n'
    >>> with open('example1.csv', 'w') as f:
    ...     f.write(s1)
    >>> a1 = read_from_filename('example1.csv')
    >>> a1
    array([[1., 2., 3.],
           [4., 5., 6.]])

    The second example has columns with different data types, so a
    one-dimensional array with a structured data type is returned.
    The tab character is used as the field delimiter.

    >>> s2 = '1.0\t10\talpha\n2.3\t25\tbeta\n4.5\t16\tgamma\n'
    >>> with open('example2.tsv', 'w') as f:
    ...     f.write(s2)
    >>> a2 = read_from_filename('example2.tsv', delimiter='\t')
    >>> a2
    array([(1. , 10, b'alpha'), (2.3, 25, b'beta'), (4.5, 16, b'gamma')],
          dtype=[('f0', '<f8'), ('f1', 'u1'), ('f2', 'S5')])

Both reference read_from_filename, which is nowhere to be found.
Maybe a rename at somepoint ?

Idea or request for content:

the function is called _read and seam to only be used in the same file in loadtxt and the imaginary_unit parameter seem to not be used in the sole call site.

If this is private, maybe move the example the the higher level function that use this one.

[ngoldbaum]

ping @seberg who added the docstrings in 1e15b43. Just looking at the commit, maybe the _read function added in that commit was originally named read_from_filename?

[krishh16]

@Carreau I would love to work on this repository

[seberg]

@ngoldbaum yeah, I think that is correct. I would agree with removing the examples (and maybe making sure that loadtxt has it, but really all that it needs as docs might be to point to loadtxt and saying that its a helper for that)