假设任务是在列表中找到最后一个满足某些条件的元素(即向后看时的第一个元素)，我得到以下数字。

Python 2:

>>> min(timeit.repeat('for i in xrange(len(xs)-1,-1,-1):\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', repeat=8))
4.6937971115112305
>>> min(timeit.repeat('for i in reversed(xrange(0, len(xs))):\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', repeat=8))
4.809093952178955
>>> min(timeit.repeat('for i, x in enumerate(reversed(xs), 1):\n    if 128 == x: break', setup='xs, n = range(256), 0', repeat=8))
4.931743860244751
>>> min(timeit.repeat('for i, x in enumerate(xs[::-1]):\n    if 128 == x: break', setup='xs, n = range(256), 0', repeat=8))
5.548468112945557
>>> min(timeit.repeat('for i in xrange(len(xs), 0, -1):\n    if 128 == xs[i - 1]: break', setup='xs, n = range(256), 0', repeat=8))
6.286104917526245
>>> min(timeit.repeat('i = len(xs)\nwhile 0 < i:\n    i -= 1\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', repeat=8))
8.384078979492188

所以，最丑的选项xrange(len(xs)-1，-1，-1)是最快的。

Python 3(不同机器):

>>> timeit.timeit('for i in range(len(xs)-1,-1,-1):\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', number=400000)
4.48873088900001
>>> timeit.timeit('for i in reversed(range(0, len(xs))):\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', number=400000)
4.540959084000008
>>> timeit.timeit('for i, x in enumerate(reversed(xs), 1):\n    if 128 == x: break', setup='xs, n = range(256), 0', number=400000)
1.9069805409999958
>>> timeit.timeit('for i, x in enumerate(xs[::-1]):\n    if 128 == x: break', setup='xs, n = range(256), 0', number=400000)
2.960720073999994
>>> timeit.timeit('for i in range(len(xs), 0, -1):\n    if 128 == xs[i - 1]: break', setup='xs, n = range(256), 0', number=400000)
5.316207007999992
>>> timeit.timeit('i = len(xs)\nwhile 0 < i:\n    i -= 1\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', number=400000)
5.802550058999998

这里，enumerate(reversed(xs)， 1)是最快的。

假设任务是在列表中找到最后一个满足某些条件的元素(即向后看时的第一个元素)，我得到以下数字。

Python 2:

>>> min(timeit.repeat('for i in xrange(len(xs)-1,-1,-1):\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', repeat=8))
4.6937971115112305
>>> min(timeit.repeat('for i in reversed(xrange(0, len(xs))):\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', repeat=8))
4.809093952178955
>>> min(timeit.repeat('for i, x in enumerate(reversed(xs), 1):\n    if 128 == x: break', setup='xs, n = range(256), 0', repeat=8))
4.931743860244751
>>> min(timeit.repeat('for i, x in enumerate(xs[::-1]):\n    if 128 == x: break', setup='xs, n = range(256), 0', repeat=8))
5.548468112945557
>>> min(timeit.repeat('for i in xrange(len(xs), 0, -1):\n    if 128 == xs[i - 1]: break', setup='xs, n = range(256), 0', repeat=8))
6.286104917526245
>>> min(timeit.repeat('i = len(xs)\nwhile 0 < i:\n    i -= 1\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', repeat=8))
8.384078979492188

所以，最丑的选项xrange(len(xs)-1，-1，-1)是最快的。

Python 3(不同机器):

>>> timeit.timeit('for i in range(len(xs)-1,-1,-1):\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', number=400000)
4.48873088900001
>>> timeit.timeit('for i in reversed(range(0, len(xs))):\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', number=400000)
4.540959084000008
>>> timeit.timeit('for i, x in enumerate(reversed(xs), 1):\n    if 128 == x: break', setup='xs, n = range(256), 0', number=400000)
1.9069805409999958
>>> timeit.timeit('for i, x in enumerate(xs[::-1]):\n    if 128 == x: break', setup='xs, n = range(256), 0', number=400000)
2.960720073999994
>>> timeit.timeit('for i in range(len(xs), 0, -1):\n    if 128 == xs[i - 1]: break', setup='xs, n = range(256), 0', number=400000)
5.316207007999992
>>> timeit.timeit('i = len(xs)\nwhile 0 < i:\n    i -= 1\n    if 128 == xs[i]: break', setup='xs, n = range(256), 0', number=400000)
5.802550058999998

这里，enumerate(reversed(xs)， 1)是最快的。

反转的内置函数很方便:

for item in reversed(sequence):

reverse的文档解释了它的局限性。

对于我必须与索引一起反向遍历序列的情况(例如，对于改变序列长度的原地修改)，我有这个函数定义了一个my codeutil模块:

from six.moves import zip as izip, range as xrange

def reversed_enumerate(sequence):
    return izip(
        reversed(xrange(len(sequence))),
        reversed(sequence),
    )

这避免了创建序列的副本。显然，相反的限制仍然适用。

使用list.reverse()，然后像往常一样迭代。

http://docs.python.org/tutorial/datastructures.html

我喜欢单行生成器方法:

((i, sequence[i]) for i in reversed(xrange(len(sequence))))

反向函数在这里很方便:

myArray = [1,2,3,4]
myArray.reverse()
for x in myArray:
    print x