How to tell a media error from an "object not found"

Last night, I was here until nearly 8:00, watching over a test of my new "enhanced SAVCHGOBJ" CL program.

Certain libraries (environment libraries for an application) on the system have corresponding IFS directories, within that application's home directory, and all of them are "of interest."

All went well until it got to the RPG Cycle program that attempts to save IFS subtrees "of interest." I found that the QCMDEXCs of the SAV command were popping up break messages at me if there wasn't an IFS directory matching the name (a fairly common occurrence). So I aborted the CL, changed the CALL to a CALL(E), and called the RPG program again.

It worked just fine, simply ignoring the cases where the name didn't have an IFS directory to save, until another problem reared its ugly head:

The tape drive needed cleaning. It probably had enough dirt to grow a bumper crop of potatoes (to use a barnyard idiom I grew up with). By about halfway through the list of possible "IFS directories of interest," the SAVs were failing from media errors. And since I was simply ignoring every exception the QCMDEXC'd SAV was throwing, the only way I knew what was going on was by signing on at another terminal, and looking at the joblog and the system operator messages.

Obviously, I'd kind of like to refine that, so that if there's a legitimate problem, the program abends instead of just sitting there, taking ten minutes each to unsuccessfully save a bunch of IFS directories, many of which only have one or two STMFs in them.

Unfortunately, by that time, I wasn't thinking all that clearly, and while I did have the sense to clean the drive, I certainly didn't have the sense to dump my joblog to a spool file. Nor, so far as I'm aware, do I have any convenient, safe way to induce media errors at will.

So how do I tell the difference between the exceptions I'm expecting, and the ones that should abend the backup?

--
JHHL