spec: mistake regarding sign of x % y

[alandonovan]

The spec currently says

The // and % operations on integers compute floored division and remainder of floored division, respectively. If the signs of the operands differ, the sign of the remainder x % y matches that of the dividend, x.

yet the Go and Java implementations currently exhibit this behavior, as does Python[23]:

Welcome to Starlark (java.starlark.net)
>> maxlong = 0x7fffffffffffffff
>> minlong = -maxlong-1
>> 1 % minlong
-9223372036854775807

What's more, Python's language reference, which we have aimed to follow here, says:

https://docs.python.org/3/reference/expressions.html#binary-arithmetic-operations

The modulo operator always yields a result with the same sign as its second operand (or zero);

I think the text of our spec is wrong, and should be made to match Python. The implementations may have incorporated the mistake into their behavior, and will need auditing. D'oh.

(Context: bazelbuild/bazel#12667)

@stepancheg

[brandjon]

Yep, all the wording I'm familiar with is that it gets the sign of the right-hand operand. You want the property that the result repeats periodically in the same way, even when x is at or about 0.