On 7/25/2013 11:53 AM, Kirk Talman wrote:
Every "bug" I have ever seen has been actually some programmer's mistake.
Then you just haven't been around long enough <g> Back in the fifties and sixties, plenty of failures occurred due to machine errors.
And some happened due to C.E. errors - in the seventies we had an extra 2MiB of slow memory on one machine, but the C.E. forgot to enable memory protection, so any errant program could overwrite foreign memory. Luckily I found the problem before our customers did.
And some happened due to microcode errors. In the eighties we got a 4341, only to find ourselves unable to log on to TSO - it would 0C4 consistently. It took me a while to track this down - the MVCK instruction would fail when a string was split over a 2KiB boundary that wasn't a 4KiB multiple. Some time later we got a new floppy, and it still failed; IBM fixed the condition of one split string, but not the case when both were split. Some time later we upgraded to a 4381 - one of the first things I ran was my MVCK test program; yep, it failed.
Gerhard Postpischil Bradford, Vermont ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
