> Not only was there discrepancies with the 'signed' zero, but IBM used to do > a core dump and bring production to a screeching halt because of a 'divide > by zero' error..... >
At least we didn't get the whole core dumped (on paper) -- worst case scenario for us students was that we'd get something like a data exception error and have our program abend with a dozen pages of dump printout or so. I was taking (and helping out) courses at a local community college, learning 370 assembly along with other stuff. I got pretty good at being able to debug stuff ;). Our computer staff made everyone put in a hook to a timing routine that would fix all those infinite loop problems - it would just abend with a systim error. FWIW, you could do the same thing (I think) in Linux with signal() and hooking onto SIGALRM, setting up code that for instance aborts after so many seconds of cpu time. Of course, with today's computers, even ones we used back then - 30 seconds of machine time is quite a lot.
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
