NewLib nano fprintf_float uppercase format

[Anfrid]

Hi, I try to format double value with useing linker key :
--specs=nano.specs -u _printf_float

If I use like this:

snprintf(buf, bufLen, fullLen, "%s%.*F", p, accuracy, value)

or that

snprintf(buf, bufLen, fullLen, "%s%.4F", p, value)

the snpintf function behaves incorrectly but if I use

snprintf(buf, bufLen, fullLen, "%s%F", p, value)

or

snprintf(buf, bufLen, fullLen, "%s%.4f", p, value)

it`s ok.

This bug I found in gcc version 4.9.3 20150303 (release) [ARM/embedded-4_9-branch revision 221220]
GNU Tools for ARM Embedded Processors)

I looked in file nano-vfpintf_float.c in newlib nano and for specificator 'g' and 'G' code
in line number 239 looks like this

if (code == 'g' || code == 'G')
{
      if (expt <= -4 || expt > pdata->prec)
 /* 'e' or 'E'. */
 code -= 2;
      else
 code = 'g';
}
...

but in line number 258 compare is present only for small 'f' char

if (code == 'f')
{
   /* 'f' fmt. */
   if (expt > 0)
     {
       pdata->size = expt;
       if (pdata->prec || pdata->flags & ALT)
  pdata->size += pdata->prec + 1;
     }
   else
     /* "0.X". */
     pdata->size = (pdata->prec || pdata->flags & ALT)
       ? pdata->prec + 2 : 1;
}

[Thomas Preud'homme (thomas-preudhomme)]

Hi Anfrid,

the 'F' specifier was introduced as part of C99 and nano targets C89 printf in order to save space. If you need the F specifier you will have to use newlib.

Best regards.