|
HP OpenVMS Systemsask the wizard |
|
The Question is: I need to create a shared object that can be accessed by Oracle to run system commands. The key here is that everything about the command must be logged. The windows implementation allows me to redirect "stdout" to a file. Then I can use "cout" or "pri ntf" to log the command and the time it began and ended. Then, using the C "system" function, I can run the command and any output from that command also gets logged to the logfile. In VMS, however, the scope of the redirection of SYS$OUTPUT does not ex tend back into the program. After the LIB$SPAWN command to redirect SYS$OUTPUT to a file, the LIB$SPAWN command exits and returns to the program. The remaining lines of code do NOT use the redirected SYS$OUTPUT. (I have attempted to do this with LIB$SE T_LOGICAL also but could not make that work) I would like to know how to redirect SYS$OUTPUT in the begining of a C program and use that redirection for the entire program (including "printf" output). Thank You, david The Answer is : Please see topic (5753) for a related discussion of I/O redirection. Use image accounting or OpenVMS security auditing, as these tools can likely log the various times of interest for you automatically. Alternatively, create subroutine which creates a process running DCL, and then pass commands to it via a mailbox, and log the required time and associated information. If you want a hack, create a small DCL command procedure that executes the designed command in the target process context -- the command can be passed via parameter -- and that also logs the desired information. (Or you could pass in the "prefix" and "suffix" commands around the target command via the mailbox.) The C system call and lib$spawn create new subprocesses, and the process context is typically entirely local to the new subprocess, unless sharing is explicitly requested. You can configure the subprocess to run entirely asynchronous to the parent, or you can have the parent wait for the subprocess to complete.
|