astksh review (was Re: Pending package status (09 May 2003))

[Karsten Fleischer]

> The only thing I am aware of is that a fix was needed to 
> inherit signal
> handler settings during vfork().  I believe that Christopher put the
> changes into CVS, but I don't know if a cygwin1.dll has been released
> with that change.

OK, I don't believe that the failed vfork() test has any other impact on
ksh.
AFAICS vfork() would eventually be used in libsfio, but since the
_lib_vfork test failed, fork() is used instead.

For the folks who don't know what we're talking about, it is this test
that fails:

#include <unistd.h>
#include <stdlib.h>
#include <signal.h>
main(argc, argv)
int     argc;
char**  argv;
{
        int     status;
        char*   cmd[3];
        if (argv[1])
                _exit(signal(SIGHUP, SIG_DFL) != SIG_IGN);
        signal(SIGHUP, SIG_IGN);
        switch (vfork())
        {
        case -1:
                _exit(1);
        case 0:
                cmd[0] = argv[0];
                cmd[1] = "test";
                cmd[2] = 0;
                execv(cmd[0], cmd);
                _exit(2);
        }
        status = 1;
        _exit(wait(&status) < 0 || status != 0);
}

Karsten