Dunno if this is the right mailinglist, please take any
discussion emerging to the right one if it isn't.

I have stumbled over the following problem which shows a discrepancy
between man waitpid, and what really happens. I include a little
program demonstrating the effect. Either I did not understand the
manpage fully (and then am happy to be taught better), or there is
a problem :) This is observed on

$ uname -a
NetBSD phaeton.entropie.net 1.6K NetBSD 1.6K (PHAETON) #3: Wed Nov 20 10:38:37 CET 2002     root@phaeton.entropie.net:/usr/src/sys/arch/i386/compile/PHAETON i386

Quote from manpage:
(RETURN VALUES)
	(...)
     If wait4(), wait3() or waitpid() returns due to a stopped or terminated
     child process, the process ID of the child is returned to the calling
     process.  If there are no children not previously awaited, -1 is returned
     with errno set to [ECHILD].  Otherwise, if WNOHANG is specified and there
     are no stopped or exited children, 0 is returned.  If an error is detect-
     ed or a caught signal aborts the call, a value of -1 is returned and
     errno is set to indicate the error.

"Otherwise (read: "if there are no childrens at all"), if WNOHANG is specified
and there are no stopped or exited children, 0 is returned."

Test program:

#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <sys/wait.h>

int main (int argc, char *argv[]) {
    int x;
    if ((waitpid(-1, &x, WNOHANG))==-1) {
        perror("waitpid"); 
        exit(EXIT_FAILURE);
    }
    return (EXIT_SUCCESS);
}

Output:

$  ./tst
waitpid: No child processes
Exit 1

Shouldn't ./tst just return EXIT_SUCCESS as waitpid should return 0 ?

regards,
-martin