On Dec 21, 2012, at 11:02 PM, Thomas Pfau <tfp...@gmail.com> wrote: > CC/DECC/NOANSI_ALIAS > /Include=[]/Standard=Relaxed_ANSI/Prefix=All/Obj=.obj/NoList/float=ieee/ieee=denorm/NAMES=(AS_IS,SHORTENED)/Def > ine=(PERL_CORE,_USE_STD_STAT=1) NUMERIC.C > > return NV_MAX; > ...................^ > In this statement, floating-point overflow occurs in evaluating the > expression "1.189731495357231765085759326628007016196477e4932l". > at line number 810 in file DISK$USERS:[pfau.FREEWARE.perl-5_16_2]numeric.c;1 > Error status %X10B91260 occurred when updating target NUMERIC.OBJ
This is a bug that goes back to <http://perl5.git.perl.org/perl.git/commit/67597c89125e7e144f9ba60f5b1fe23b951286d8?f=numeric.c> eleven years ago. We misspelled _IEEE_FP (which is what the compiler defines) as __IEEE_FP, with an extra leading underscore. So code that was intended as a workaround in overflow handling for folks who had D_FLOAT or G_FLOAT has in fact been always included on VMS. It compiles without error except when configuring with 64-bit integers but without long doubles. This should fix it: --- numeric.c;-0 Thu Oct 25 21:44:15 2012 +++ numeric.c Sat Dec 22 08:45:46 2012 @@ -803,7 +803,7 @@ S_mulexp10(NV value, I32 exponent) * a hammer. Therefore we need to catch potential overflows before * it's too late. */ -#if ((defined(VMS) && !defined(__IEEE_FP)) || defined(_UNICOS)) && defined(NV_MAX_10_EXP) +#if ((defined(VMS) && !defined(_IEEE_FP)) || defined(_UNICOS)) && defined(NV_MAX_10_EXP) STMT_START { const NV exp_v = log10(value); if (exponent >= NV_MAX_10_EXP || exponent + exp_v >= NV_MAX_10_EXP) ________________________________________ Craig A. Berry mailto:craigbe...@mac.com "... getting out of a sonnet is much more difficult than getting in." Brad Leithauser
