Hi Peter, Peter Klein wrote: > Hallo, > > yesterday I took autogen 5.9.3, compiled and installed it on my > Ultra 5 Spac Stations with the 64 bit version of OpenBSD 3.4 and > 3.6. On both machines the test reported 4 failures, to be reported. > > Please see the log file of the tests attached and let me know, > if you need additional information.
Thank you. I'll combine the replies. :) I've fixed the stuff for which I could see causes. Please try: http://autogen.sourceforge.net/data/autogen-5.9.4pre7.tar.gz It fixes those, but not the two silent failures (see below). Thank you again! Regards, Bruce Problem 1 + ./nested -s stumble, foo, <bar type=integer>1234</bar> able + sed / nested:/s/ 0x.*/ 0xXXXXXXXX/ + > nested-res1.out + cmp -s nested-res1.base nested-res1.out + diff nested-res1.base nested-res1.out + failure 5c5 < bar -- integer: 1234 --- > bar -- integer: 5299989643264 $ printf '0x%X\n' 5299989643264 0x4D200000000 ~ $ printf '0x%X\n' 1234 0x4D2 I'm going out on a limb here: You're running on a system where sizeof(long) is 8. What is sizeof(int)? Byte order? My development system is x86-64, but sizeof(int) is still 4. Let's see the code. Hmmm. Rats. Quite the think-o: $ cvs diff -up nested.c ! /usr/bin/cvs diff ... Index: nested.c =================================================================== RCS file: /home/cvs/cvsroot/autogen/autoopts/nested.c,v retrieving revision 4.24 diff -u -p -r4.24 nested.c --- nested.c 13 Nov 2007 05:49:26 -0000 4.24 +++ nested.c 17 Nov 2007 05:34:45 -0000 @@ -219,12 +219,12 @@ addNumberValue( void** pp, char const* p dataLen--; pzValue++; } if (dataLen == 0) - pNV->v.boolVal = 0; + pNV->v.longVal = 0; else - pNV->v.boolVal = atoi( pzValue ); + pNV->v.longVal = strtol(pzValue, 0, 0); pNV->valType = OPARG_TYPE_NUMERIC; - pNV->pzName = (char*)(pNV + 1); + pNV->pzName = (char*)(pNV + 1); memcpy( pNV->pzName, pzName, nameLen ); pNV->pzName[ nameLen ] = NUL; addArgListEntry( pp, pNV ); Problem 2 looks familiar: + diff -c config.out config.res + failure *** config.out Fri Nov 16 10:09:49 2007 --- config.res Fri Nov 16 10:09:49 2007 *************** *** 4,10 **** over everything. alpha -- no value beta -- string: , gamma ! zzyzx -- integer: 180388626432 YES!! alpha -- no value beta -- string: , gamma --- 4,10 ---- over everything. alpha -- no value beta -- string: , gamma ! zzyzx -- integer: 42 YES!! alpha -- no value beta -- string: , gamma $ printf '0x%X\n' 180388626432 0x2A00000000 ~ $ printf '0x%X\n' 42 0x2A Problem 3 is different: + compile_getopt 1 /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/agen5/autogen -L/gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts getopt.def /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts/test/testdir/getopt-inc/ccV10620.o: In function `main': /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts/test/testdir/getopt-inc/.ag-T25515/test_getopt.c:349: undefined reference to `optionProcess' /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts/test/testdir/getopt-inc/.ag-T25515/test_getopt.c:352: undefined reference to `optionPutShell' /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts/test/testdir/getopt-inc/ccV10620.o:(.data+0x98): undefined reference to `optionUsage' /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts/test/testdir/getopt-inc/ccV10620.o:(.data+0x370): undefined reference to `optionPrintVersion' /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts/test/testdir/getopt-inc/ccV10620.o:(.data+0x440): undefined reference to `optionPagedUsage' collect2: ld returned 1 exit status Killing AutoGen: cannot compile test_getopt.c in /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts/test/testdir/getopt-inc/.ag-T25515 AutoGen aborting on signal 15 (Terminated) in state EMITTING processing template /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts/usage.tpl on line 123 for function EXPR (11) Abort (core dumped) FAILURE: could not generate getopt-test_getopt.c For this, I need the stuff in /gnu/gnu/built/OpenBSD/3.4/sparc64/autogen/autogen-5.9.3/autoopts/test/FAILURES That should include the compiler error messages. Problem 4: is very different I don't see any error messages for the enums.test test. If you roll up that FAILURES directory, there will be some logs for it. If there are no files starting with "enums" in that directory, try this: gmake verbose TESTS=enums.test and send the contents of "testdir". ====================================================SOLARIS======= Problem 1: BSD/GNU-ism: + /gnu/gnu/built/SunOS/5.9/gnu/autogen/autogen-5.9.3/agen5/autogen -L/gnu/gnu/built/SunOS/5.9/gnu/autogen/autogen-5.9.3/autoopts getopt.def /bin/sh: mktemp: not found Never mind. That lack is accommodated and ``2>/dev/null'' did not make the message go away. + gcc -std=gnu99 -DHAVE_CONFIG_H -g -O2 -I/gnu/gnu/built/SunOS/5.9/gnu/autogen/autogen-5.9.3 -I/gnu/gnu/built/SunOS/5.9/gnu/autogen/autogen-5.9.3/agen5 -I/gnu/gnu/built/SunOS/5.9/gnu/autogen/autogen-5.9.3/autoopts -I/usr/gnu/include -I /gnu/gnu/built/SunOS/5.9/gnu/autogen/autogen-5.9.3/autoopts/test/testdir/getopt-inc -c getopt2-getopt.c getopt2-getopt.c:21:20: getopt.h: No such file or directory getopt2-getopt.c:35: error: elements of array `a_long_opts' have incomplete type getopt2-getopt.c:36: warning: excess elements in struct initializer "getopt_long" requires glibc and "getopt.h". I'll have to adjust the test. It will be disabled unless /usr/include/getopt.h exists. Problem 2: another silent failure: FAIL: nls.test I'll need the same treatment as above with the enums.test. Thanks. ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Autogen-users mailing list Autogen-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/autogen-users
