How to see the size of the incoming floating point number?

问题

The user writes a number to the input, it is stored in a string. How can I check if this number is included in size in the float type or does it need a double?

回答1:

Unless your floating point numbers are huge or extremely small, i.e. out of the range spanning -3.4E38 to 3.4E38, a float 32 will store anything you throw at it in terms of size but not accuracy. As such, the real issue is how many significant digits you need in order to minimize rounding errors. I recommend you to read https://www.itu.dk/~sestoft/bachelor/IEEE754_article.pdf

If you are not limited by disk space or memory, then just go for float 64.

回答2:

How can I check if this number is included in size in the float type or does it need a double?

Numbers encoded as strings offer limitless possibilities. Finite float and double are limited in range and precision.

Note that float is a subset of double.

The set of values of the type float is a subset of the set of values of the type double; C17dr § 6.2.5 10

Range

The range of double typically well exceeds that of float.

Precision

Typical float and double are a 2^N * a dyadic rational: some integer/some-power-of-two. So conversion from string to floating point involves some rounding. E.g. 0.1 is not typically exactly representable as float nor as a double.

This implies most inexact conversions, even if in float range, will have a closer answer as double than float.

---

To meet OP's goal, I'd suggest converting the string to both and test the conversion results.

int float_or_double_range(const char *s) {
  char *endptr;
  errno = 0;
  double d = strtod(s, &endptr);
  if (s == endptr) return 'n';  // Neither
  if (errno == ERANGE) return 'd';

  errno = 0;
  double f = strtof(s, &endptr);
  if (s == endptr) return 'd';
  if (errno == ERANGE) return 'd';

  if (d == f) return 'f'; // encodable as float and double
  return 'd';
}

---

Notes:

Recall that the correctness of FP strto...() functions are subject to quality of implementation issues and that they themselves may not provide the best answers in all cases.

To find if the converted string value is the same as a double and float, I recommend against converting the string to double and then the double to float. That involves double rounding and introduces errors in corner cases.

回答3:

This answer is only for positive float´s but it might help you out:

A 32-bit float (8 byte/ Single precision) as defined by IEEE 754 has the largest positive float number of 3.40282 x 10^38, the smallest positive float number is 1.17549 x 10^-38.

Use strtod() to convert the number in the string to a double. This is needed because you actually don´t know if the number is already double or not.

Then check if the number is within the range provided above.

If it is, allocate a float. If not, continue to use the double object.

This way is a bit muddy, because you already allocate a double, then choosing for either use the double used as buffer before or allocate another float. Nonetheless, this is beneficial if you allocate for example an array based of the number of the string. Furthermore, you also have the option to dynamically allocate the buffer double object and free() it after its use.

---

A much simpler way would be to just choose a double from start. So ensure that this process is really required. Unless it isn´t an explicit prohibition to use a double, just use a double.

This will save you a lot of time and effort and is with that also the safest way to go.

来源：https://stackoverflow.com/questions/61709217/how-to-see-the-size-of-the-incoming-floating-point-number