TODO
    - bugfix:
      - fix the v0.018 t/02*.t roundoff bug: it was overflowing 32bit on {ivsize}=4 systems;
        new method should not
      - testing on my 32b linux box, found failures in t/01*.t suite; needs to be investigated
    - long-term: implement more functions from IEEE754-2007, especially
      + IEEE754 5.4.2 => 5.12
        ☐ <format>-convertFormat
        ☐ <format>-convertFromDecimalCharacter
            # converts a string decimal number (ie, output of to_dec_floatingpoint)
            #   into a <format>-floating point value
        ☑ <format>-convertToDecimalCharacter(source, conversionSpecification)
            # alias for to_dec_floatingpoint()
            # conversionSpecification is supposed to indicate formatting
            #       might want to use an optional integer which indicates how many places
            #       after the fractional-point to display (default to 16 => %.16f for to_dec_floatingpoint)
      + IEEE754 5.4.3 => 5.12
        ☐ <format>-convertFromHexCharacter
            # converts a string hexadecimal number (ie, output of to_hex_floatingpoint)
            #   into a <format>-floating point value
        ☑ <format>-convertToHexCharacter(source, conversionSpecification)
            # alias for to_hex_floatingpoint()
            #       I originally thought these were aliases for my rawhex754 functions... but no
            # conversionSpecification is supposed to indicate formatting
            #       might want to use an optional integer which indicates how many places
            #       after the fractional-point to display (default to 13 => %13.13s for to_hex_floatingpoint)
      + valid <format> = qw(binary32 binary64 binary128)

    - long-term: add in _32bit and _128bit versions
      - <format> = qw(binary32 binary64 binary128)
      - 128b = [1 sign][15 exp][112 frac], e=exp-16383

    - ALWAYS: look for new bug reports:
      <https://rt.cpan.org/Dist/Display.html?Queue=Data-IEEE754-Tools>
      <https://github.com/pryrt/Data-IEEE754-Tools/issues>