>The issue Peter is facing is that a certain amount of confusion ensues when a >particular piece of software notices the 00A coming from a COBOL program, >because the overflow bit has become set on and has somehow avoided means to >turn it off before a C/C++ program returns to COBOL.
Reduced to the minimum: Another software's condition handler is getting control *before* LE's. This one does *not* ignore the program check 0A, but closes, backs out and disconnects, then percolates to LE. LE in turn knows it should ignore the program check 0A for Cobol, and returns control to the application code. The application later fails with random symptoms because of the "closing" actions. -- Peter Hunkeler ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
