Re: Weird INFDS problem

Barbara Morris wrote:

James Lampert wrote:

I've got a strange issue in which one particular OPM program (OPM RPG, if it makes any difference), when running on one particular box (the one on which it was compiled), is unable to access the INFDS of its display file, and blows up with a "pointer not set" error if it attempts to do so.
...

It sounds like storage corruption. The OPM RPG compiler defines the INFDS as based on an internal pointer. If that pointer gets corrupted, then any access to the INFDS would fail with a pointer not set.

If you pass parameters to some other program, maybe that other program did change, to expect larger parameters.

Assuming I'm right, you might be able to narrow down where the corruption is happening by adding several statements that use the INFDS into your program. Code a few COMP statements like the one below into your program, and run it. When it fails on one of them, add the COMP statement at every statement between the previous good one and the failing one. Run the program again, and you should be able to see exactly where the problem starts.

C INFDS COMP *BLANKS 10

Fascinating idea. Although it bothers me a bit: I've used this "checkpoint" concept many times, and yet somehow, I didn't think of applying it here.

Maybe it has something to do with the fact that all week, I've been dreaming in weird (but pleasant) flashbacks to my fall vacation: mostly in immersive guided tours that were pastiches of Colonial Williamsburg, Walt Disney World, and Kennedy Space Center. (For some reason, though, Washington, DC hasn't figured in those dreams, even though it represented more than three times more of this particular vacation than Kennedy Space Center did.)

--
JHHL