[Python-ideas] More useful slices

[Chris Barker]

On Mon, Feb 2, 2015 at 10:00 AM, Skip Montanaro <skip.montanaro at gmail.com>
wrote:

> As I indicated in an earlier response, there is some performance value
> to replacing the range function call with this (or other) syntactic
> sugar.


It's my impression that adding stuff like this primarily for performance is
nothing on the list ;-)

But you could get the same effect by making range a keyword -- how often
does anyone re-define it?

And if you allow variables:

range(i,j,k)  or (i:j:k)

Then any optimizer needs to to run-time type checking, or type inference,
or what have you anyway, so checking if the name range is the range object
seems like a small issue.

FWIW, Cython does turn range calls into simple C loops:

for i in range(10):

becomes:

for (__pyx_t_1 = 0; __pyx_t_1 < 10; __pyx_t_1+=1) {

...
}

In that case, the cython "language" doesn't allow re-defining of "range"

And it will do this if either literals or variables that have been
type-def'd as integer types are passed in as arguments.

What difference this really makes to performance I have no idea -- you'd
have to have something pretty trivial in that loop to notice it.

-Chris





> While exceedingly rare, there is nothing to prevent a
> programmer from redefining the range builtin function or inserting a
> different version of range() in the local or global scopes:
>
>     def my_range(*args):
>         # completely ignore args, returning something weird...
>         return [1, "a", None, 2]
>
>     import __builtin__
>     __builtin__.range = my_range
>
> If you have a for loop which uses range():
>
>     for i in range(27):
>         do something interesting with i
>
> optimizers like PyPy which aim to be precisely compatible with CPython
> semantics must look up "range" every time it occurs and decide if it's
> the real builtin range function, in which case it can emit/execute
> machine code which is something like the C for loop:
>
>     for (i=start; i<stop; i+=step) {
>         do something interesting with i
>     }
>
> or not, in which case it has to call whatever range is, then respond
> with its list of (arbtrary) Python objects.
>
> Syntactic sugar would lock down the standard behavior.
>
> Skip
> _______________________________________________
> Python-ideas mailing list
> Python-ideas at python.org
> https://mail.python.org/mailman/listinfo/python-ideas
> Code of Conduct: http://python.org/psf/codeofconduct/
>



-- 

Christopher Barker, Ph.D.
Oceanographer

Emergency Response Division
NOAA/NOS/OR&R            (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chris.Barker at noaa.gov
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-ideas/attachments/20150202/37fad5d4/attachment.html>