据说在 python 下，某个情况下 "i += x" 不等于 "i = i + x"？

Question

某个不愿理透露姓名的大大跟我说：

小伙子啊，你的代码有潜在的 bug 啊，i += x 会粗问题的。我思考了良久，都不得要领。

那么请问，什么情况下 "i += x" 不等于 "i = i + x"？

Answer 1

Look at the comparison of the two pieces of code below:

Code 1:

>>> l1=range(3)
>>> l2=l1
>>> l2+=[4]
>>> l1
[0, 1, 2, 4]
>>> l2
[0, 1, 2, 4]

Code 2:

>>> l2=l1
>>> l2=l2+[4]
>>> l1
[0, 1, 2]
>>> l2
[0, 1, 2, 4]

See here for a detailed explanation: http://stackoverflow.com/questions/2347265/why-does-behave-unexpectedly-on-lists

Answer 2

Obviously, just like what the 2nd floor said.
Code 1 l2 = l1, l2 += [4], all operations are l1, similar to pointers, referencing Shenma, if you don’t understand it, just think about it.
Code 2 l2 = l1, l2 = l2 + [4], this is obviously a reassignment of l2. You can write l3 = l2 + [4], l2 += [4]. Then you will know the result.
You can learn about python’s deep copy and shallow copy

There is a very good answer in stackoverflow on the second floor. Just use id() to check the memory address

>>> l = []
>>> id(l)
13043192
>>> l += [3]
>>> id(l)
13043192
>>> l = l + [3]
>>> id(l)
13059216

---

Don’t worry too much, just look at the memory address and it will be clear at a glance.
id is a built-in function of python, what's id? ...

《'id' is a bad variable name in Python》

id() is a fundamental built-in:
Help on built-in function id in module builtin:

id(...)

    id(object) -> integer

    Return the identity of an object.  This is guaranteed to be unique among
    simultaneously existing objects.  (Hint: it's the object's memory
    address.)

Answer 3

What I can think of is that some classes implement __iadd__ and behave differently from __add__. List's += is an example. Ordinary + will return a new list instance, while += is Directly manipulate the list itself.

If you can confirm that i and x are two numbers, saying that i += x will cause problems is a bit nitpicking and dogmatism.

Answer 4

Mainly look at i, if i is an expression
In i+=x, i is only calculated once
i = i+x i needs to be calculated 2 times

Then, if there are modification variables in the i expression, the values ​​of the two i in i = i+x may be different

a[i++] += 5         // i 执行一次求值
a[i++] = a[i++] + 5 // i 执行两次求值

This example should be common in various grammar books

Answer 5

The above is for reference types, there should be no difference for value types.

Answer 6

The core issue is to prevent side effects. Make your code behave consistently.

Answer 7

The poster here has some misunderstandings, i += x, this is a kind of syntax sugar, what does it really look like after expansion?
In C language, the true form of i += x is i = i+x. Here i is reassigned, so the pointer address of i will change.
But this is not the case in Python. The true form of i += x is i.extend(x), where the pointer address of i has not changed, only the value of i has been changed.

refer: http://stackoverflow.com/questions/2347265/why-does-behave-unexpectedly-on-lists
thank to: @WKPlus

Answer 8

What Floor 4 said is so right. There is a chance that the MD5 values ​​of two different files are the same. The question is, have you ever encountered them?