William Heymann wrote:
> This is a pretty small change but I would suggest xrange instead of range.
Good point! Since I was calling range() during the benchmark, it was
timed too. Switching to xrange() will mean less overhead.
I re-ran this benchmark (again):
s = ""
for i in range(100000):
s += "a"
The result:
Python 2.5 release: 30.1s
Python 2.5 locally built: 30.4s
Python 2.5 concat: 3.95s
Improvement: *669%*! (1/7.7 the time.)
So xrange(10000000) was adding between 0.4s and 0.9s overhead, and
losing it makes my approach look even better. Thanks!
/larry/
--
http://mail.python.org/mailman/listinfo/python-list
