Re: frexp man page: FLT_RADIX vs. 2

[Matthew House <mattlloydhouse@gmail.com>]

On Tue, Jan 2, 2024 at 8:49 PM Alejandro Colomar <alx@kernel.org> wrote:
> The value of that macro is defined to be 2; it can't have any other
> value.
>
> ISO C defines it in
> <http://port70.net/~nsz/c/c11/n1570.html#5.2.4.2.2p11>.
>
> POSIX defines it in
> <https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/float.h.html>.
>
> Maybe the definition of frexp() by ISO C and POSIX should be changed to
> define it in terms of FLT_RADIX instead of 2.
>
> Have a lovely day,
> Alex

Every version of ISO C says (emphasis mine):

  The values given in the following list shall be replaced by constant
  expressions with implementation-defined values that are *greater or
  equal* in magnitude (absolute value) to those shown, with the same sign:

  -- radix of exponent representation, b

     FLT_RADIX                        2

And POSIX defines it similarly. So FLT_RADIX can be greater than 2 (unless
the implementation defines __STDC_IEC_559__), it just can't be any less.
Indeed, the subsequent EXAMPLE 1 in ISO C sets FLT_RADIX to 16.

Thus, since frexp(3) is defined to return a power of 2, interpreting it as
a power of FLT_RADIX is incorrect in the general case.

Thank you,
Matthew House