Re: [Mingw-users] putenv(...) and system(...)

[Oscar F. <of...@wa...>]

Daniel Taupin <ta...@wa...> writes:

>  If, inside a program XXX, I code
> 
>    putenv("set environt_var=its_value");
> 
> and then
> 
>    system("some-program.exe");
> 
> will the program some-program.exe have "environt_var" set to the
> specified value?

Changes introduced by 'putenv' affects child processes. Now, I'm not
100% sure if processes created by 'system' are considered child
processes or independent ones. Let's perform a quick check:

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

int main(int argc, char *argv[]) {
  char command[100];
  command[0] = 0;

  if(argc > 1) {
    printf("FOO value on secondary process: %s\n", getenv("FOO"));
  }
  else {
    printf("Setting FOO to BAR on initial process.\n");
    putenv("FOO=BAR");
    strcat(command, argv[0]); /* This program's name */
    strcat(command, " dummy");
    printf("Calling system...\n");
    system(command);
  }
}


See the output of the above program.
 
> Second question: will the variable "environt_var" stay at "its_value"
> when XXX has terminated.

A child process can't change the environment variables of its parent
process. This includes the case when the parent process is a shell and
the child process is a command issued to the shell.
 
> Note that I ask that quesiton to know whether that coding could have a
> behaviour different of what happens when using "system(...)" to invoque
> a BATCH file which performs the two operations.
> 
> Last question: is that valid whatever the system: NT, 9x, 2k, XP ?

It should. *nix systems too.

-- 
Oscar