What does the Python Ellipsis object do?

Question

While idly surfing the namespace I noticed an odd looking object called "Ellipsis", it does not seem to be or do anything special, but it's a globally available builtin.

After a search I found that it is used in some obscure variant of the slicing syntax by Numpy and Scipy... but almost nothing else.

Was this object added to the language specifically to support Numpy + Scipy? Does Ellipsis have any generic meaning or use at all?

D:\workspace\numpy>python
Python 2.4.4 (#71, Oct 18 2006, 08:34:43) [MSC v.1310 32 bit (Intel)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> Ellipsis
Ellipsis

Answer 1

You can use Ellipsis yourself, in custom slicing situations like numpy has done, but it has no usage in any builtin class.

I don't know if it was added specifically for use in numpy, but I certainly haven't seen it used elsewhere.

See also: http://stackoverflow.com/questions/118370/how-do-you-use-the-ellipsis-slicing-syntax-in-python

Answer 2

From the Python documentation:

This object is used by extended slice notation (see the Python Reference Manual). It supports no special operations. There is exactly one ellipsis object, named Ellipsis (a built-in name).

Answer 3

This came up in another question recently. I'll elaborate on my answer from there:

Ellipsis is an object that can appear in slice notation. For example:

myList[1:2, ..., 0]

Its interpretation is purely up to whatever implements the __getitem__ function and sees Ellipsis objects there, but its main (and intended) use in in the numeric python extension, which adds a multidementional array type. Since there are more than one dimensions, slicing becomes more complex than just a start and stop index; it is useful to be able to slice in multiple dimentions as well. eg, given a 4x4 array, the top left area would be defined by the slice "[:2,:2]"

>>> a
array([[ 1,  2,  3,  4],
       [ 5,  6,  7,  8],
       [ 9, 10, 11, 12],
       [13, 14, 15, 16]])

>>> a[:2,:2]  # top left
array([[1, 2],
       [5, 6]])

Extending this further, Ellipsis is used here to indicate a placeholder for the rest of the array dimensions not specified. Think of it as indicating the full slice [:] for all the dimentions in the gap it is placed, so for a 3d array, a[...,0] is the same as a[:,:,0] and for 4d, a[:,:,:,0]. Similarly a[0,...,0] is a[0,:,:,0] (with however many colons in the middle make up the full number of dimensions in the array)

Interestingly, in python3, the Ellipsis literal (...) is usable outside the slice syntax, so you can actually write:

>>> ...
Ellipsis

Other than the various numeric types, no, I don't think its used. As far as I'm aware, it was added purely for numpy use and has no core support other than providing the object and corresponding syntax. The object being there didn't require this, but the literal "..." support for slices did.