Danny Smith wrote: > Or > am I missing a more fundamental reason for the limited range of targets > supporting DFP?
Enabling it in gcc would just enable the compiler parts, but the decimal float C extension proposed standard includes a lot more than that, there's a whole checklist of stuff required of the libc/libm that AFAIK is only currently implemented by glibc (and thus the current limitation to linux): http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1176.pdf - float.h support for macros like DEC{32,64,128}_MANT_DIG, DEC{32,64,128}_{MIN,MAX}_EXP, et al. - math.h support for HUGE_VAL_D{32,64,128}, DEC_INFINITY, DEC_NAN, quantized{32,64,128}, et al. - rounding defines FE_DEC_{TOWARDZERO,TONEAREST,...} and fe_dec_{get,set}round() - stdlib.h support for strtod{32,64,128}() - wchar.h support for wcstod{32,64,128}() - %H, %D, %DD printf specifiers - %DF, %DD, %DL scanf specifiers Brian