[Numpy-discussion] (Late) summary of PEP-225 discussion at Scipy

Fernando Perez fperez.net@gmail....
Wed Oct 22 18:14:56 CDT 2008


Hi

Please keep your replies on list.  It's important that this thread in
particular is publicly archived, since it may serve as the reference
for further discussions.

On Wed, Oct 22, 2008 at 3:37 PM, Sebastien Binet <binet@cern.ch> wrote:
> Fernando,
>
>> much delayed, but here it is, finally.  The doc regarding our
>> discussion about PEP 225 is attached, and I'm keeping a public copy
>> for viewing convenience (with html) here:
>>
>> https://cirl.berkeley.edu/fperez/static/numpy-pep225/
>>
>> Note that, as indicated in the link above, the real doc is in bzr, so
>> you can check it out to generate patches or a  branch (the preferred
>> formats for big changes).
>>
>> This is just a first cut, going from memory and notes.  I'd appreciate
>> any feedback, corrections, etc.
>
> what about the D-language-like syntax I proposed which has the advantage of not
> introducing gazillions of new operators (but just stating that the operation
> will be broadcasted/vectorized) ?
>
> # create array of doubles
> a = ...
> # add 42 to all elements
> a[] += 42.
> # apply a function
> b = sin(a[])
> # or this?
> b = sin[](a)
>
> c = a[] + b[]

Regardless of its merits in D, I'd bet this has little chance of
flying in python.  The pep is simply about adding some special methods
with matching operator form, inline with existing python syntactic
tradition.  This suggestion is a pretty radical departure from
Python's existing style.

Having said that, I'm not Guido and I could be 100% wrong here, so
I'll most certainly put your comments in when I update the document
later.

>> I'm giving a talk at the BayPiggies
>> group Nov 13 about SciPy
> count me in ;)
>
> cheers,
> sebastien.
> --
> ###################################
> # Dr. Sebastien Binet
> # Lawrence Berkeley National Lab.
> # 1 Cyclotron Road
> # Berkeley, CA 94720
> ###################################
>
>

Cheers,

f


More information about the Numpy-discussion mailing list