Re: More on 1.0.0 BETA; ftp mirrors

lance@fox.com
Sun, 2 Jan 94 17:20:59 PST

> Congratulations, Lance, for being the first to report your porting
> experiences! Some quick replies:
>
> Re your hacks for struct timeval: there is some code in mytime.h and
> in the configure script that tries to cope with exactly this situation
> -- apparently it has failed. Could you mail me the config.h header
> file generated by configure?

here is my config.h file:
/* Include/config.h. Generated automatically by configure. */
/* NOTE: config.h.in is converted into config.h by the configure
script in the toplevel directory.

On non-UNIX systems, manually copy config.h.in to config.h, and
edit the latter to reflect the actual configuration of your system.

Then arrange that the symbol HAVE_CONFIG_H is defined during
compilation (usually by passing an argument of the form
`-DHAVE_CONFIG_H' to the compiler, but this is necessarily
system-dependent). */

/* Types which have no traditional name -- edit the definition if necessary */

#define RETSIGTYPE void /* int or void: return of signal handlers */

/* Types which are often defined in <sys/types.h> -- either define as
some variant of int or leave undefined. Uncomment a definition if
your <sys/types.h> does not define the type */

/* #define mode_t int */
/* #define off_t long */
/* #define pid_t int */
/* #define size_t unsigned */
/* #define uid_t int */
/* #define gid_t int */

/* Feature test symbols -- either define as 1 or leave undefined */

/* symbol name: #define as 1 if: */

#define STDC_HEADERS 1 /* the standard C header files exist
(in particular, <stdlib.h>,
<stdarg.h>, <string.h> and <float.h>) */

/* #undef HAVE_DLFCN_H */ /* <dlfcn.h> exists */
#define HAVE_SIGNAL_H 1 /* <signal.h> exists */
#define HAVE_STDARG_H 1 /* <stdarg.h> exists (else need <varargs.h>) */
#define HAVE_STDLIB_H 1 /* <stdlib.h> exists */
#define HAVE_UNISTD_H 1 /* <unistd.h> exists */
#define HAVE_UTIME_H 1 /* <utime.h> exists */

#define HAVE_SYS_PARAM_H 1 /* <sys/param.h> exists */
#define HAVE_SYS_SELECT_H 1 /* <sys/select.h> exists */
#define HAVE_SYS_TIMES_H 1 /* <sys/times.h> exists */
/* #undef HAVE_SYS_TIME_H */ /* <sys/time.h> exists */
#define HAVE_SYS_UTSNAME_H 1 /* <sys.utsname.h> exists */

#define TIME_WITH_SYS_TIME 1 /* <sys/time.h> and <time.h> can be included
together */

/* #undef HAVE_TM_ZONE */ /* struct tm has a tm_zone member */
#define HAVE_TZNAME 1 /* extern char *tzname[] exists */

#define HAVE_CLOCK 1 /* clock() exists */
#define HAVE_FTIME 1 /* ftime() exists */
#define HAVE_GETPGRP 1 /* getpgrp() exists */
#define HAVE_GETTIMEOFDAY 1 /* gettimeofday() exists */
#define HAVE_LSTAT 1 /* lstat() exists */
#define HAVE_PROTOTYPES 1 /* the compiler understands prototypes */
#define HAVE_READLINK 1 /* readlink() exists */
#define HAVE_SELECT 1 /* select() exists */
#define HAVE_SETPGID 1 /* setpgid() exists */
#define HAVE_SETPGRP 1 /* setpgrp() exists */
#define HAVE_SETSID 1 /* setsid() exists */
#define HAVE_SYMLINK 1 /* symlink() exists */
/* #undef HAVE_SIGINTERRUPT */ /* siginterrupt() exists */
#define HAVE_TCGETPGRP 1 /* tcgetpgrp() exists */
#define HAVE_TCSETPGRP 1 /* tcsetpgrp() exists */
#define HAVE_TIMES 1 /* times() exists */
#define HAVE_UNAME 1 /* uname() exists */
#define HAVE_WAITPID 1 /* waitpid() exists */

#define GETPGRP_HAVE_ARG 1 /* getpgrp() must be called as getpgrp(0)
(and setpgrp() as setpgrp(0, 0)) */

/* #undef WITH_READLINE */ /* GNU readline() should be used */
/* #undef USE_THREAD */ /* Build in thread support */
/* #undef SOLARIS */ /* This is SOLARIS 2.x */

> Re AF_UNIX: Wouldn't testing for the presence of AF_UNIX in the header
> file be sufficient to determine whether to include it in the cases or
> not? I hate to build dependencies on a particular version of a
> particular system into the source...

Nope.. AF_UNIX is defined in the header files eventhough it is not
possible to use.. also, AF_UNIX is defined, but no 'un.h' header file.
Thus I do not know of any other way to do it except for building for
just SCO ODT.

> Re test_grammar failure: my guess would be that your strtoul() doesn't
> check for overflow. As a temporary solution, add strtoul.o to OBJS in
> Makefile(.in) and see if that solves the matter; if so, I'll see if I
> can come up with a test for this in the configure script. (Or maybe I
> should just always use my "own" strtoul?)
>

That fixed the problem... Thank you for the suggestion....

--
Lance Ellinghouse                lance@fox.com