On Sat, Aug 2, 2008 at 11:05 AM, Roland Mainz <[EMAIL PROTECTED]> wrote: > Dennis Clarke wrote: >> On Fri, Aug 1, 2008 at 11:27 PM, Ming Kin Lai <[EMAIL PROTECTED]> wrote: >> > I have a feeling that Sun has abandoned maintaining Solaris, and shifted >> > its resources to OpenSolaris. Well, that's good news to people here, I >> > guess. >> > So, can I discuss things about Solaris 10 here? How about something about >> > Solaris 10's C library? Can I ask a question why a certain library >> > function does not work properly? Can I report a bug here? Currently >> > there does not appear to be a way to report Solaris' bugs in Sun's website >> > anymore. >> > >> >> just a guess, are you referring to an issue with strtod(3C) in >> C99/SUSv3 mode NOT being able to handle hexadecimal strings ? > > Erm... the hexfloat format, e.g. ... > -- snip -- > $ printf "%a\n" 0.1 > 0x1.999999999999999999999999999ap-04 > -- snip -- > ... is AFAIK part of C99 standard and should work on Solaris. If it does > not work then this is a bug...
It doesn't and it is because of a bug in the m4 makefiles ... I think. Let me drag out the email about this from a few days ago. ah yes .. from the m4 project : > > According to Dennis Clarke on 7/29/2008 7:48 PM: > | not too sure what the issue is here. > > | test-strtod.c:738: assertion failed > ... > | FAIL: test-strtod > > The issue is well-known. Thank you for the quick reply. I did a little bit of research and have seen many other posts about this. I should do some research before I file a bug report next time. > Your libc is buggy when it comes to C99 strtod(3) compliance I am surprised to see that in Solaris 10. The Sun folks tend to be fascists about compliance to specs and standards and that is why I love them. see : http://www.blastwave.org/man/s10/standards_5.html where I see : Compilation A POSIX.1 (1988-1996)-, XPG4-, SUS-, or SUSv2-conforming implementation must include an ANSI X3.159-1989 (ANSI C Language) standard-conforming compilation system and the cc and c89 utilities. A POSIX.1-2001- or SUSv3-conforming implementation must include an ISO/IEC 99899:1999 (1999 ISO C Language) standard-conforming compilation system and the c99 utility. Solaris 10 was tested with the cc, c89, and c99 utilities and the compilation environment provided by Sun Studio C Compiler 5.6. At the very bottom of that page I also see : Similar SUSv3-conforming LP64 applications should be built with command lines of the form: c99 $(getconf POSIX_V6_LP64_OFF64_CFLAGS) -D_XOPEN_SOURCE=600 \ $(getconf POSIX_V6_LP64_OFF64_LDFLAGS) foo.c -o foo \ $(getconf POSIX_V6_LP64_OFF64_LIBS) -lxnet SUSv3 c99 _XOPEN_SOURCE=600 OKay ... so I'll go back and rebuild m4 1.4.11 with this : bash-3.2$ uname -a SunOS mercury 5.10 Generic_127128-11 i86pc i386 i86pc bash-3.2$ c99 -V cc: Sun C 5.9 SunOS_i386 Patch 124868-01 2007/07/12 usage: c99 [ options] files. Use 'c99 -flags' for details Also, I am going to have a hard look at the Sunsolve bug reports and patches released and see if there is anything filed about this. If there is a real concern about C99 compliance in libc then the Sun folks will know about it. > but the configure check in m4 1.4.11 didn't check > for all the known bugs in various libc's. You can use './configure > gl_cv_func_strtod_works=no' to work around it; or use the latest snapshot: > > http://home.comcast.net/~ericblake/m4-1.5.89a.31-d9538d.tar.gz > http://home.comcast.net/~ericblake/m4-1.5.89a.31-d9538d.tar.gz.asc > http://home.comcast.net/~ericblake/m4-1.5.89a.31-d9538d.tar.bz2 > http://home.comcast.net/~ericblake/m4-1.5.89a.31-d9538d.tar.bz2.asc > http://home.comcast.net/~ericblake/m4-1.5.89a.31-d9538d.tar.lzma > http://home.comcast.net/~ericblake/m4-1.5.89a.31-d9538d.tar.lzma.asc Oh well, thank you very much. I'll drag down one of those promptly and then see what happens after a build with c99 and after libc is patched. Oh, and I'll circle around and give you my progress report of that is okay. Dennis Clarke http://www.blastwave.org ----------- I found that the m4 1.4.11 release will fail the strtod test every time unless I specify that we should skip it, of course. The newer releases ( or CVS ) does not fail but also will not link because of some problems with makefiles. I was able to get a nice 64-bit m4 but I don't know if I trust it : $ file /export/home/dclarke/local/bin/m4 /export/home/dclarke/local/bin/m4: ELF 64-bit LSB executable AMD64 Version 1 [SSE2 SSE FXSR CMOV FPU], dynamically linked, not stripped $ m4 --version m4 (GNU M4) 1.4.11 Copyright (C) 2008 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Written by Rene' Seindal. I had to manually compile and link the strtod code or it would fail and I'd have an assertion message every time. >From test-strtod.c in the m4 tests directory : /* Hex. */ { const char input[] = "0xa"; char *ptr; double result; errno = 0; result = strtod (input, &ptr); ASSERT (result == 10.0); /* NetBSD 3.0, OpenBSD 4.0, AIX 5.1, HP-UX 11.11, IRIX 6.5, OSF/1 5.1, Solaris 10, mingw */ ASSERT (ptr == input + 3); /* NetBSD 3.0, OpenBSD 4.0, AIX 5.1, HP-UX 11.11, IRIX 6.5, OSF/1 5.1, Solaris 10, mingw */ ASSERT (errno == 0); } that would fail every time. Let me rebuild : test-strtod.c:738: assertion failed test-strtod.c:739: assertion failed test-strtod.c:748: assertion failed test-strtod.c:749: assertion failed test-strtod.c:758: assertion failed test-strtod.c:759: assertion failed test-strtod.c:768: assertion failed test-strtod.c:769: assertion failed test-strtod.c:778: assertion failed test-strtod.c:779: assertion failed test-strtod.c:788: assertion failed test-strtod.c:789: assertion failed okay .. the code that fails ... $ cat -n test-strtod.c | head -792 | tail -63 | cut -c1-44 730 731 /* Hex. */ 732 { 733 const char input[] = "0xa"; 734 char *ptr; 735 double result; 736 errno = 0; 737 result = strtod (input, &ptr); 738 ASSERT (result == 10.0); 739 ASSERT (ptr == input + 3); 740 ASSERT (errno == 0); 741 } 742 { 743 const char input[] = "0XA"; 744 char *ptr; 745 double result; 746 errno = 0; 747 result = strtod (input, &ptr); 748 ASSERT (result == 10.0); 749 ASSERT (ptr == input + 3); 750 ASSERT (errno == 0); 751 } 752 { 753 const char input[] = "0x1p"; 754 char *ptr; 755 double result; 756 errno = 0; 757 result = strtod (input, &ptr); 758 ASSERT (result == 1.0); 759 ASSERT (ptr == input + 3); 760 ASSERT (errno == 0); 761 } 762 { 763 const char input[] = "0x1p+"; 764 char *ptr; 765 double result; 766 errno = 0; 767 result = strtod (input, &ptr); 768 ASSERT (result == 1.0); 769 ASSERT (ptr == input + 3); 770 ASSERT (errno == 0); 771 } 772 { 773 const char input[] = "0x1p+1"; 774 char *ptr; 775 double result; 776 errno = 0; 777 result = strtod (input, &ptr); 778 ASSERT (result == 2.0); 779 ASSERT (ptr == input + 6); 780 ASSERT (errno == 0); 781 } 782 { 783 const char input[] = "0x1p+1a"; 784 char *ptr; 785 double result; 786 errno = 0; 787 result = strtod (input, &ptr); 788 ASSERT (result == 2.0); 789 ASSERT (ptr == input + 6); 790 ASSERT (errno == 0); 791 } 792 Here comes a compile with cc : $ cc -g -m64 -xO1 -Qy -Xa -xbuiltin=%none -xdebugformat=dwarf -xmodel=medium -xnolibmil -xnolibmopt -xregs=no%frameptr -xs -xstrconst -xtarget=opteron -xtemp=/tmp/dclarke -I../lib -H -o test-strtod_assert_fail test-strtod.c ../lib/config.h ../lib/stdlib.h /usr/include/stdlib.h /usr/include/iso/stdlib_iso.h /usr/include/sys/feature_tests.h /usr/include/sys/ccompile.h /usr/include/sys/isa_defs.h /usr/include/iso/stdlib_c99.h /usr/include/sys/wait.h /usr/include/sys/types.h /usr/include/sys/machtypes.h /usr/include/ia32/sys/machtypes.h /usr/include/sys/int_types.h /usr/include/sys/select.h /usr/include/sys/time_impl.h /usr/include/sys/time.h /usr/include/sys/types.h /usr/include/time.h /usr/include/iso/time_iso.h /usr/include/sys/select.h /usr/include/sys/resource.h /usr/include/sys/siginfo.h /usr/include/sys/machsig.h /usr/include/vm/faultcode.h /usr/include/sys/procset.h /usr/include/sys/signal.h /usr/include/sys/iso/signal_iso.h /usr/include/sys/unistd.h ../lib/unistd.h /usr/include/unistd.h ../lib/stdlib.h /usr/include/errno.h /usr/include/sys/errno.h /usr/include/float.h ../lib/math.h /usr/include/math.h /usr/include/iso/math_iso.h /usr/include/iso/math_c99.h /usr/include/floatingpoint.h /usr/include/stdio_tag.h /usr/include/sys/ieeefp.h ../lib/stdio.h /usr/include/stdio.h /usr/include/iso/stdio_iso.h /usr/include/sys/va_list.h /usr/include/stdio_impl.h /usr/include/iso/stdio_c99.h /usr/include/stdarg.h /usr/include/iso/stdarg_iso.h /usr/include/sys/va_impl.h /usr/include/iso/stdarg_c99.h /usr/include/stddef.h /usr/include/iso/stddef_iso.h ../lib/string.h /usr/include/string.h /usr/include/iso/string_iso.h $ file test-strtod_assert_fail test-strtod_assert_fail: ELF 64-bit LSB executable AMD64 Version 1 [SSE2 SSE FXSR CMOV FPU], dynamically linked, not stripped $ ./test-strtod_assert_fail test-strtod.c:738: assertion failed test-strtod.c:739: assertion failed test-strtod.c:748: assertion failed test-strtod.c:749: assertion failed test-strtod.c:758: assertion failed test-strtod.c:759: assertion failed test-strtod.c:768: assertion failed test-strtod.c:769: assertion failed test-strtod.c:778: assertion failed test-strtod.c:779: assertion failed test-strtod.c:788: assertion failed test-strtod.c:789: assertion failed Here we go with c99 : $ c99 -g -m64 -xO1 -Qy -Xa -xbuiltin=%none -xdebugformat=dwarf -xmodel=medium -xnolibmil -xnolibmopt -xregs=no%frameptr -xs -xstrconst -xtarget=opteron -xtemp=/tmp/dclarke -I../lib -H -o test-strtod_assert_fail test-strtod.c ../lib/config.h ../lib/stdlib.h /usr/include/stdlib.h /usr/include/iso/stdlib_iso.h /usr/include/sys/feature_tests.h /usr/include/sys/ccompile.h /usr/include/sys/isa_defs.h /usr/include/iso/stdlib_c99.h /usr/include/sys/wait.h /usr/include/sys/types.h /usr/include/sys/machtypes.h /usr/include/ia32/sys/machtypes.h /usr/include/sys/int_types.h /usr/include/sys/select.h /usr/include/sys/time_impl.h /usr/include/sys/time.h /usr/include/sys/types.h /usr/include/time.h /usr/include/iso/time_iso.h /usr/include/sys/select.h /usr/include/sys/resource.h /usr/include/sys/siginfo.h /usr/include/sys/machsig.h /usr/include/vm/faultcode.h /usr/include/sys/procset.h /usr/include/sys/signal.h /usr/include/sys/iso/signal_iso.h /usr/include/sys/unistd.h ../lib/unistd.h /usr/include/unistd.h ../lib/stdlib.h /usr/include/errno.h /usr/include/sys/errno.h /usr/include/float.h ../lib/math.h /usr/include/math.h /usr/include/iso/math_iso.h /usr/include/iso/math_c99.h /usr/include/floatingpoint.h /usr/include/stdio_tag.h /usr/include/sys/ieeefp.h ../lib/stdio.h /usr/include/stdio.h /usr/include/iso/stdio_iso.h /usr/include/sys/va_list.h /usr/include/stdio_impl.h /usr/include/iso/stdio_c99.h /usr/include/stdarg.h /usr/include/iso/stdarg_iso.h /usr/include/sys/va_impl.h /usr/include/iso/stdarg_c99.h /usr/include/stddef.h /usr/include/iso/stddef_iso.h ../lib/string.h /usr/include/string.h /usr/include/iso/string_iso.h $ file test-strtod_assert_fail test-strtod_assert_fail: ELF 64-bit LSB executable AMD64 Version 1 [SSE2 SSE FXSR CMOV FPU], dynamically linked, not stripped $ ./test-strtod_assert_fail works fine with c99 :-) ( that rhymes ) so ... its a funny looking failure in the m4 code because the strtod manpage says that "In C99/SUSv3 mode, hexadecimal strings are also recognized." and thus the stuff fails when I use plain jane[1] cc but works with c99. Dennis [1] I knew a Jane and she was not plain _______________________________________________ opensolaris-code mailing list opensolaris-code@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
