C++20 是否为 "overflow" 的有符号整数很好地定义了左移?

Does C++20 well-define left shift for signed integers that "overflow"?

在目前的C++标准草案中,左移运算符定义如下[expr.shift]:

The value of E1 << E2 is the unique value congruent to E1×2^E2 modulo 2^N, where N is the width of the type of the result.

考虑具有 32 位的 int E1 = 2^31-1 = 2'147'483'647E2 = 1int。则有无限个数与E1×2^E2 = 4'294'967'2942^N = 2^32同余,即所有数4'294'967'294 + k×2^32其中k是一个任意整数。例如 4'294'967'294 (k=0) 或 -2 (k=-1).

我不明白标准中这些数字中的唯一值是什么意思。是指可以用结果数据类型表示的唯一值吗?然后,我假设结果定义为 -2。这个解释正确吗?

在 C++20 之前,定义有所不同,这种情况会导致未定义的行为。我想这个变化与负符号整数的强制性 2 的补码表示有关。

事实上,现在不再要求E1为非负数。因此,似乎 -1 << 1 被定义为 -2。是吗?

所以我们知道:

E1 = 2147483647
E2 = 1
N = sizeof(int) * CHAR_BIT = 4 * 8 = 32

让我们计算 E1×2^E2 modulo 2^Nmodulo 是除法的余数):

x = E1×2^E2 mod 2^N = 2147483647 * 2 ^ 1 mod 4294967296 = 4294967294 mod 4294967296 = 4294967294

然后我们去here:

For each value x of a signed integer type, the value of the corresponding unsigned integer type congruent to x modulo 2 N has the same value of corresponding bits in its value representation.

而且我认为我们还需要:

The base-2 representation of a value of signed integer type is the base-2 representation of the congruent value of the corresponding unsigned integer type.

这意味着,对于 signed intx = 4294967294 等于 x = -2。所以结果将是 -2.

It therefore seems that -1 << 1 is defined as -2. Is it right as well?

 (signed)-1 << 1 = 
 4294967295 << 1 = 
 4294967295 * 2 ^ 1 mod 4294967296 = 
 8589934590 mod 4294967296 = 
 4294967294 = 
 (signed)-2

Does it mean the unique value that can be represented by the resulting data type

是的。与 E1×2^E22^N 一致的一组数是无限的,但在任何大小 2^N 的区间中只有一个值,因此只有一个值可以表示为宽度的整数类型N.

如果我们查看 "p0907R1 Signed Integers are Two’s Complement" proposal,我们会发现一个与 "unique representation" 相似的短语,这使这一点更加清楚:

Conversion from signed to unsigned is always well-defined: the result is the unique value of the destination type that is congruent to the source integer modulo 2N.

Then, I suppose the result is defined as -2. Is this interpretation correct?

在 x64 上等效的 asm 指令是 shlx(逻辑左移)

I suppose the change is related to the mandatory 2-complement representation of negative signed integers.

正确。与无符号类型的情况一样,现在也有符号类型,它们在数学上表示等价 类(好吧,我不清楚这在多大程度上是正确的,因为看起来他们仍然希望在溢出时保留一些 UB 案例)。