Long integers, xrange and number theory

[Delaney, Timothy]

> From: Paul Rubin [mailto:phr-n2002b at NOSPAMnightsong.com.avaya.com]
> 
> martin at v.loewis.de (Martin v. Löwis) writes:
> > > xrange would be the different story, since it involves a proxy
> > > object that supports a sequence protocol.
> > 
> > Indeed. That proxy object would need to change, making it 
> inefficient
> > even for small integers.
> 
> No, with small integer arguments the xrange function could just
> construct the same proxy object it does now.  It would only make a new
> kind of object if long ints were involved.

Just a quick implementation ... we can do better ;)

from __future__ import generators

def _xrange (start, stop, step):

    while 1:

        if step > 0 and start > stop:
            return
        
        if step < 0 and start < stop:
            return
        
        yield start
        start += step        

def xrange (start, stop=None, step=None, xrange=xrange):

    if stop is None:
        stop = start
        start = 0
    
    if step is None:
        step = 1

    try:
        return xrange(start, stop, step)
    except OverflowError:
        return _xrange(start, stop, step)

if __name__ == '__main__':

    for i in xrange(10):
        print i

    print

    for i in xrange(1000000000000000L, 1000000000000010L):
        print i

Tim Delaney