Yeeh, I was expecting something like that. The only reason to use map() at all is for improving the performance.
That is lost when using list comprehensions (as far as I know). So, this is *no* option for larger jobs.
Don't believe anything you hear right away, especially not when it comes to performance, especially not wrt. Python.
[EMAIL PROTECTED]:~$ python -m timeit -s "items=['']*10000;import string" "map(string.upper, items)"
100 loops, best of 3: 6.32 msec per loop
[EMAIL PROTECTED]:~$ python -m timeit -s "items=['']*10000;import string" "[s.upper for s in items]"
100 loops, best of 3: 2.22 msec per loop
So using map is *no* option for larger jobs.
Of course that statement is also false. Performance prediction is very difficult, and you cannot imply much from this benchmark. In other cases, list comprehension may be slower than map. More likely, for real (i.e. non-empty) strings, the cost of .upper will make the precise implementation of the loop irrelevant for performance reasons.
Regards, Martin -- http://mail.python.org/mailman/listinfo/python-list
