Re: ORACLE RAC CRASHES ORBITZ

> Never heard of a C compiler crash that interrupted production

Well - In 1983 I had a 300 line COBOL program that whenever you compiled it would cause the mainframe to abend. Seriously. We could never work out why - change one character of the program, and the problem would go away. Used to love giving the Sys Admin a hard time by threatening to compile it unless he ran my batch job NOW !

My other favourite from that era was Get Mains for memory allocation. The MF OS had a fixed upper limit to how many it stored in a mapping table - (256 I believe). Along came #257, and it would quite happily start poking the new single byte map reference into the location in the memory stack. One customer had this problem where the 8 bits overwritten coincidently where also the static boolean constants for Yes/No and True/False (which suddenly became set to their opposite values) in their program logic.

So every test for True and False, Yes and No in the code suddenly started acting backwards, after the series of code executed that cased the 1 too many get mains - in a 12 million line program. Took two of us a week to work that one out. Received on Mon Jul 28 2003 - 21:47:14 CDT