Performance of f-string vs string.join

Question

one = 'one'
two = 'two'
three = 'three'

Are the 2 equivalent in time complexity?

result_1 = f'{one}{two}{three}'
result_2 = ''.join([one, two, three])

I know that join works in linear time. But is the same true also for f-strings? With f-strings, the number of strings being concatenated is always a constant, so I guess one could say it's constant time complexity, but I was more curious to know how the concatenation is implemented and how it would fare with a large number of strings.

Answer 1

The performance of these two approaches may vary between releases of Python and could also depend on your hardware.

In Python 3.11.3 on macOS 13.4 (3 GHz 10-Core Intel Xeon W) the f-string implementation is faster than str.join()

from timeit import timeit

ONE = 'one'
TWO = 'two'
THREE = 'three'

def func1(one, two, three):
    return f'{one}{two}{three}'

def func2(one, two, three):
    return ''.join([one,two,three])

for func in func1, func2:
    print(func.__name__, timeit(lambda: func(ONE, TWO, THREE), number=2_000_000))

Output:

func1 0.2694316829999934
func2 0.33222394099993835

Note:

This result will be due in part to the fact that func2 has to construct a list. This can be eliminated by re-writing it as:

def func2(*args):
    return ''.join(args)

...which timeit reports as:

func2 0.30138257099997645