The mi_process_exec() function

The mi_process_exec() function forks a process and executes the commands in the argv array. The rules for the contents of the argv array are as follows:

• The first element must be an executable program or script.

  You must provide the full path name of this program or script. A single environment variable is allowed as a prefix using the same rules as the mi_file_open() function.

  To include environment variables in the command path name, use the following syntax:

  $ENV_VAR

  To use the PATH environment variable, you must execute the command through a shell (such as /bin/sh). However, make sure that the command is also executable:

  /bin/sh -c cmd

  If the command is a script, it must be executable and start with a shell identifier (such as “#!/bin/sh”).

• The last element must be the keyword NULL.

For example, the following code fragment executes the UNIX™ touch command on a file called somefile:

char *argv[5];
argv[0] = "/usr/ucb/touch";
argv[1] = "somefile"; 
argv[2] = NULL;
mi_process_exec(argv);

The following code fragment executes the same command but uses the Bourne shell (/bin/sh) to execute it:

argv[0] = "/bin/sh"; 
argv[1] = "-c"; 
argv[2] = "touch"; 
argv[3] = "somefile"; 
argv[4] = NULL; 
mi_process_exec(argv);

The new process inherits its execution environment from the parent server-initialization process (the operating-system process that runs the oninit utility or its equivalent). The user and group identifiers are those of the application user of the session, and the working directory is as follows:

$ONEDB_HOME/bin