Issue 21176: Implement matrix multiplication operator (PEP 465)

[Nathaniel Smith at numpy-discussion]

Guido just formally accepted PEP 465:
  https://mail.python.org/pipermail/python-dev/2014-April/133819.html
  http://legacy.python.org/dev/peps/pep-0465/#implementation-details

Yay.

The next step is to implement it, in CPython and in numpy. I have time
to advise on this, but not to do it myself, so, any volunteers? Ever
wanted to hack on the interpreter itself, with BDFL guarantee your
patch will be accepted (if correct)?

The todo list for CPython is here:
    http://legacy.python.org/dev/peps/pep-0465/#implementation-details
There's one open question which is where the type slots should be
added. I'd just add them to PyNumberMethods and then if someone
objects during patch review it can be changed.

Here's a first patch. It works, but needs more tests (associativity, precedence, __rmatmul__, etc...) and also docs. I can work on that tomorrow.

Here are changes to the operator module...

I've got to the point where I can do

>>> import ast
>>> ast.dump(ast.parse('a @ b'))
"Module(body=[Expr(value=BinOp(left=Name(id='a', ctx=Load()), op=MatMult(), right=Name(id='b', ctx=Load())))])"


I'll post a link to my bitbucket clone shortly.

Here is a nearly complete patch...

Wow! That was quick. And I am still fighting with bitbucket.  Maybe I should stop duplicating the effort at this point.

On Mon, Apr 7, 2014, at 22:09, Alexander Belopolsky wrote:
> 
> Alexander Belopolsky added the comment:
> 
> Wow! That was quick. And I am still fighting with bitbucket.  Maybe I
> should stop duplicating the effort at this point.

Sorry about that. I only saw the first message, where Nathaniel invites
an implementation. I didn't realize that was a direct quote from the
mailing list (not by you) and didn't notice you also assigned it to
yourself.

Thanks for stepping in.  From a quick look at your patch I don't see anything that I would do much differently.

I noticed some whitespace issues in Include/token.h:

-#define AT              49	
-#define RARROW          50
-#define ELLIPSIS        51
+#define AT              49
+#define ATEQUAL		50
+#define RARROW          51
+#define ELLIPSIS        52

(Are you sure you want to re-enumerate RARROW and ELLIPSIS? Why not just give ATEQUAL the value of 52?)

+   .. versionadded:: 3.4

Are you planning to use the time machine? :-)

On Mon, Apr 7, 2014, at 22:23, Alexander Belopolsky wrote:
> 
> Alexander Belopolsky added the comment:
> 
> Thanks for stepping in.  From a quick look at your patch I don't see
> anything that I would do much differently.
> 
> I noticed some whitespace issues in Include/token.h:
> 
> -#define AT              49     
> -#define RARROW          50
> -#define ELLIPSIS        51
> +#define AT              49
> +#define ATEQUAL                50
> +#define RARROW          51
> +#define ELLIPSIS        52
> 
> (Are you sure you want to re-enumerate RARROW and ELLIPSIS? Why not just
> give ATEQUAL the value of 52?)

I thought it would be nicer if ATEQUAL was right after AT.

On Mon, Apr 7, 2014, at 22:25, Alexander Belopolsky wrote:
> 
> Alexander Belopolsky added the comment:
> 
> +   .. versionadded:: 3.4
> 
> Are you planning to use the time machine? :-)

Good catch. :)

Here's my latest and greatest version. I'll finish up after some sleep.

Here's what I consider a complete patch.

I think this patch is fairly straightforward, and I don't want it to rot, so here we go...

New changeset c553d8f72d65 by Benjamin Peterson in branch 'default':
PEP 465: a dedicated infix operator for matrix multiplication (closes #21176)
http://hg.python.org/cpython/rev/c553d8f72d65

nosy: + python-dev
messages: + msg215865

resolution: fixed
stage: needs patch -> resolved

messages: + msg215740

messages: + msg215732

messages: + msg215730

nosy: + benjamin.peterson
messages: + msg215726

keywords: + patch