I read about a very clever student compiler implementation somewhere. It initialized all integer fields in storage to x'80000000'. Then it followed every storage read with an LCR -- so at the cost of one fast instruction per storage read, it forced an overflow exception on any read of uninitialized storage. The only additional cost was the loss of -2147483648 as a valid integer value.
The bad parity trick does not lose an integer value or cost any time, but it does require a degree of hardware manipulation not possible on Z. Charles -----Original Message----- From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of Seymour J Metz Sent: Thursday, March 3, 2022 7:24 AM To: [email protected] Subject: Re: S0C4 pic 4 Subject to the proviso that once you've found the proximate cause of the S0C4 as some anomaly, you'll probably still need to track down the cause of that anomaly. There used to be a student compiler for FORTRAN IV that initialized storage to bad parity and put out an "uninitialized" message if you read that storage before you set it. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
