Re: Debug getting lost?

Vern Hamberg wrote:

Ending the job that was using a replaced program will break the linkage - it's the job that resolves the address of program objects - well, something in the activation group - maybe one can say that ending an AG would do it. Is that true?

When debugging gets weird, I almost always solve it by ending the job.

Hey, Vern.

Sure, but ending a job always ends the program instructions that were executing in that job. Why would "SEP" code be any different? What is a 'job' outside of the code executing within it? (And some system infrastructure...)

Again, if a "SEP" is partly implemented via some small piece of code in the *PGM object that notifies debug services when the program starts, then it makes some sense that QRPLOBJ could be where the 'real' target of a SEP is after a recompile. It could explain why a SEP still exists but doesn't get activated when the new copy runs without a fairly complete reestablishment.

It would also indicate an obvious way to test it. Just run the program from QRPLOBJ. It might need to be recovered to its original library and even named back to it's original name if SEP code kept track.

If nothing else, it might eliminate consideration of one possibility.

Tom Liotta