Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!clyde.concordia.ca!uunet!samsung!caesar.cs.montana.edu!ogicse!plains!overby From: overby@plains.UUCP (Glen Overby) Newsgroups: sci.space Subject: Re: The time thread that won't die. Summary: some of us will be partying Message-ID: <3034@plains.UUCP> Date: 23 Dec 89 20:21:29 GMT References: <12649@maven.u.washington.edu> Reply-To: overby@plains.UUCP (Glen Overby) Organization: North Dakota State University, Fargo Lines: 30 In article <12649@maven.u.washington.edu> games@maven.u.washington.edu (Games Wizard) writes: >There will be >>NO<< PARTY for software engineers on Jan 1 2000, because at >that point all of the old bad programmers will have to figure out how to fix >all the programs with 89 instead of 1989 as dates in data. For them >at least the old milennia will end, and a new one ( a kinder gentler >more user-friendly one ) will begin. Ah, but you are forgetting that not everybody uses the last two digits of the year! Some systems, like Unix, the Mac, and DOS keep time relative to an epoch (Midnight Jan 1, 1970 for Unix). Not that the OS' internal representation can prevent an application program from making the mistake you describe... Those systems will blow up at a later time, when the time quantum exceeds what can be represented by an unsigned integer. So far, sizeof(int) has increased faster than time. Some programmers who practice other unclean methods will also have problems when the time exceeds a positive integer, and their time will become negative. For the most part, these problems are left for a future generation of softwear engineers to work out. >"and the master programmer spake : Now my son, you are truly enlightened." "There is always a better way." -- Thomas Edison -- Glen Overby uunet!plains!overby (UUCP) ncoverby@ndsuvax, overby@plains (Bitnet)