Same function but different names with different set of default arguments

[Steven D'Aprano]

On Thu, 21 Jan 2016 06:30 pm, Paulo da Silva wrote:

> Hi all.
> 
> What is the fastest implementation of the following code?

Let's find out. Here are three different ways to do it:

def g(p):
    return

def f1(p=3):  # argument with a default
    return g(p)

def f2():  # no argument at all
    return g(3)

from functools import partial
f3 = partial(g, 3)


# setup timing code
from timeit import Timer
t1 = Timer("f1()", "from __main__ import f1")
t2 = Timer("f2()", "from __main__ import f2")
t3 = Timer("f3()", "from __main__ import f3")


Now let's see how long they take. This is using Python 2.7 on my computer.


py> min(t1.repeat(repeat=7))
0.43099188804626465
py> min(t1.repeat(repeat=7))
0.4344518184661865
py> min(t1.repeat(repeat=7))
0.42992687225341797



py> min(t2.repeat(repeat=7))
0.43400001525878906
py> min(t2.repeat(repeat=7))
0.432689905166626
py> min(t2.repeat(repeat=7))
0.43417787551879883


py> min(t3.repeat(repeat=7))
0.281879186630249
py> min(t3.repeat(repeat=7))
0.27957892417907715
py> min(t3.repeat(repeat=7))
0.28043699264526367


There's no visible difference between the first and second method. The third
method, using functools.partial, is considerably faster, BUT remember that
this only effects the time it takes to call the function g(). If g()
actually does any work, the time spent doing the work will *far* outweigh
the overhead of calling the function.




-- 
Steven