A pythonic merge sort implementation

mergesort.py

""" Straightforward approach, shaving sorted elements off of the beginning of the partitions """ 
def merge(l_left, l_right):
    l_sorted = []
    while l_left and l_right:
        if l_left[0] <= l_right[0]:
            l_sorted.append(l_left.pop(0))
        else: 
            l_sorted.append(l_right.pop(0))
    # Extend return variable by list with remaining entries
    return l_sorted + l_left + l_right

""" Build sorted partitions in reverse to avoid having to pop(0) and decrease runtime """
def merge_rev(l_left, l_right):
    l_sorted = []
    while l_left and l_right:
        if l_left[-1] >= l_right[-1]:
            l_sorted.append(l_left.pop())
        else:
            l_sorted.append(l_right.pop())
    l_sorted.reverse()
    return l_left + l_right + l_sorted

def merge_sort(l, merge_f):
    len_l = len(l)
    if len_l <= 1: return l
    mid = len_l//2
    return merge_f(merge_sort(l[:mid], merge_f), merge_sort(l[mid:], merge_f))

import random
li = random.sample(range(50000), 50000)

print('merge    ', end=' ')
%timeit merge_sort(li, merge)

print('merge_rev', end=' ')
%timeit merge_sort(li, merge_rev)

print('Timsort  ', end=' ')
%timeit sorted(li))

"""
OUTPUT: ----------------------------------------------------------------------

merge     634 ms ± 6.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)
merge_rev 306 ms ± 1.44 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)
Timsort   16.6 ms ± 351 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
"""