3 ms·

Shouldn't you write return (long)x + y; to ensure the addition happens after the conversion to long?

• rustybolt 5y ago
  Yes, this is essentially the case I'm making: if this implementation of binary search is buggy, then so is the addition operator in C (and many related languages).
  • cxr 5y ago
    They're not comparable. Trying to compare them suggests that you don't understand what the issue is with the `(low + high) / 2` case, or you don't want to acknowledge it. Simple addition will fail if the result of the expression should fall outside the range of integers expressible for a given integer width—because it can't fit. That's not insightful. It's facile. The criticism of the naive midpoint computation is not facile. The problem with using `(low + high) / 2` to compute the midpoint (instead of `(high - low) / 2 + low`) is that it will give invalid results even when the data type from the signature is wide enough that the expected result can fit.
  • thfuran 5y ago
    No, it's just slightly unintuitive. That's not the same thing.
• ufo 5y ago
  That might not be enough. On 64-bit linux, both int and long are 32 bits.
  • adrian_b 5y ago
    You meant on 64-bit Windows. On most 64-bit UNIX-like systems, long is the same with long long.
    • mark-r 5y ago
      Microsoft are the kings of backward compatibility. Since long has been 32 bits for the last 20 years, it shall remain so until the end of time. Linux compilers are free to be a little more pragmatic. The standards are flexible enough to allow either behavior.