🚀 go-pugleaf

seebs@solutions.solon.com (Peter Seebach) writes:

> How are we to test this?

I'm afraid there's no way to test whether a floating value exceeds
INT_MAX that works on all conceivable conforming implementations.

You might try comparing ((long double) f) to ((long double) INT_MAX);
this comparison works on all implementations I know about.

> how can one determine the maximum number of digits one will need
> to print an integer?

If you don't need the exact maximum, but can settle for a conservative
estimate that may be slightly too large, then you can use the following idea,
which I first saw in code written by Arthur David Olson.  Again, it's
not guaranteed on all conceivable conforming implementations,
but it works on all implementations I know about.

/*
** 302 / 1000 is log10(2.0) rounded up.
** Subtract one for the sign bit;
** add one for integer division truncation;
** add one more for a minus sign.
*/
#define INT_STRLEN_MAXIMUM(type) \
	((sizeof(type) * CHAR_BIT - 1) * 302 / 1000 + 2)