Re: [Chicken-users] Return value of `system'

[Kon Lovett]

On Mar 14, 2007, at 2:55 PM, Zbigniew wrote:

> Exit values should be distinguishable from signals (and even from
> signals+core dumps).  Either use the C macros, or assume bits 15-8 are
> the exit return code, bit 7 is the core dump, and bits 6-0 are the
> signal number (with 127 being shell exec failure).  It's the same as
> in Perl unless Chicken does something weird with the return from
> system; granted I haven't looked at the code.  Hopefully Chicken's
> (system) returns the value of system(), not errno.  system() never
> sets errno to my knowledge.

From "runtime.c":

C_regparm C_word C_fcall C_execute_shell_command(C_word string)
{
  int n = C_header_size(string);
  char *buf = buffer;

  if(n >= STRING_BUFFER_SIZE) {
    if((buf = (char *)C_malloc(n + 1)) == NULL)
      barf(C_OUT_OF_MEMORY_ERROR, "system");
  }

  C_memcpy(buf, ((C_SCHEME_BLOCK *)string)->data, n);
  buf[ n ] = '\0';

  n = C_system(buf);

  if(buf != buffer) C_free(buf);

  if(n == -1)
    return C_fix(errno);

#ifdef C_NONUNIX
  return C_fix(n);
#else
  return C_fix(WIFEXITED(n) ? WEXITSTATUS(n) : (WIFSIGNALED(n) ?  
WTERMSIG(n) : WSTOPSIG(n)));
#endif
}

I don't always talk thru my hat ;-)

Best Wishes,
Kon

> On 3/14/07, Kon Lovett <address@hidden> wrote:
> > IMHO this means the return value is almost useless w/o some hoops;
> > i.e. if 'errno' == the return value the return value is error value,
> > but differentiating an exit-status from a signal-code is probably
> > impossible.