THE END of the first unix gigasecond IS NEAR

Is anyone concerned about the rollover of time_t from 9 digits to 10? This is happening at 2001-09-09 01:46:40 GMT, which is next Saturday night, or evening in the US. Check your code for S1G bugs, such as ordering timestamps using textual rather than numeric comparisons, or assuming that time_t columns require no more than 9 characters. I have set an alarm but I plan to celebrate the dawn of the new gigasecond quietly. I once made the mistake of alerting my (non-geek) friends that a leap second had just occurred ...

Scott Harrington wrote:
Is anyone concerned about the rollover of time_t from 9 digits to 10? This is happening at 2001-09-09 01:46:40 GMT, which is next Saturday night, or evening in the US. Check your code for S1G bugs, such as ordering timestamps using textual rather than numeric comparisons, or assuming that time_t columns require no more than 9 characters.
I have set an alarm but I plan to celebrate the dawn of the new gigasecond quietly. I once made the mistake of alerting my (non-geek) friends that a leap second had just occurred ...
in late 1998, I received a message concerning a number of upcoming dates that were likely to cause a problem. I am including the message for your amusement. Some of the things that are suggested as problems seem to be rather bizarre (the Jan 1, 1999 problem for example). I can believe some of the problems might have occurred or will occur but they just sound strange. I wonder where some of this information came from. As for the gigasecond, there is a reference to use of 999,999,999 as an end of file marker and the problems that it will cause. And with all of the things that he included, he left out the 2100 problem...... -- Martin Smoot Network Storage Solutions 703-834-2242 msmoot@nssolutions.com www.nssolutions.com --------------------------------------------------------------------------------------- As I'm sure you all heard, computer life as we know it will go a little nuts on Jan. 1, 2000 (unless, of course, you use a Mac): ------------------ Jan. 1, 2000, isn't only 'doomsdate' Steve Woodward / Newhouse News Service Jan. 1, 2000, is The Big One, kids. By now, you've heard that many of the world's computers will roll the date clock forward from "99" to "00" with potentially disastrous consequences. Year 2000 authorities prophesy problems as minor as erroneous overdue notices from the library and as major as a failure of the nation's power grid. But that isn't the only computer "doomsdate" looming. A slew of lesser-known dates also could wreak technological havoc. So brace yourself. The first date to dread -- Jan. 1, 1999 -- is fast approaching. Jan. 1, 1999: The one-year-look-ahead problem Not every computer counts forward like you and me. Some look down the road one entire year and count backward to determine the date. (Please don't ask why.) On Jan. 1, 1999, some will look forward one year and see "00." Like humans, the computers may balk at having to count backward from 00. Jan. 1, 1999, to Dec. 31, 2002: The euro currency problem We all know that the year 2000 problem is the biggest software project in history. But many Americans are unaware that programmers throughout the world are also at work on the second biggest software project in history: converting the currencies of 11 European nations into a single currency called the euro. Banks and financial institutions will begin transacting business in euros on Jan. 1, 1999, although the actual bank notes won't be issued until Jan. 1, 2001. The introduction of the euro is to continue through the year 2002. There's no direct link between the euro project and the Y2K project, but the massive size of the simultaneous projects will soon take most of the world's available programmers. Aug. 21, 1999: The GPS rollover problem The world's 24 global positioning satellites record time by counting the weeks that have passed since their launch in 1980. The weeks fill up a counter much like the odometer on your car. But like your odometer, the counter rolls over to 0000 when it's full. At midnight on Aug. 21, 1999, the counter will be full. Equipment that uses the GPS signals may malfunction. Sept. 9, 1999: The 9999 end-of-file problem Many computers have been programmed to recognize 9999 as an "end-of-file" command. Perhaps some computers will conclude, quite logically, that a date of 9/9/99 means it's the end of all time. Oct. 1, 1999: The federal fiscal year 2000 problem Big Daddy rolls its clock forward Oct. 1, 1999. As of that date, the federal government officially enters its 2000 budget year. Every federal function will be affected, from defense to Medicare to payments on the federal debt. Jan. 4, 2000: The first-working-day-of-the-year problem Year 2000 begins on a Saturday. Corporate America will switch on most of its desktop computers Tuesday, Jan. 4, after a long holiday weekend. Boot up and hang on to your morning mochas. Feb. 29, 2000: The Year 2000 leap year problem, Part I Most programmers know the rules for calculating leap years: Any year evenly divisible by four is a leap year, except years that also are divisible by 100. So 1996 is a leap year, but 2000 isn't -- er, right? Well, there's a third, lesser-known rule that cancels the first two: Any year divisible by 400 is a leap year, including -- you guessed it -- 2000. The question is: How many programmers know that rule? Dec. 31, 2000: The Year 2000 leap year problem, Part II Some computers work by counting the number of days in the year. If they aren't programmed to know that 2000 is a leap year, the machines will be bewildered when they reach Dec. 31, 2000, the seemingly impossible 366th day of the year. Sept. 8, 2001: The Unix end-of-file problem Unix is the "other" major operating system, a set of instructions that, like Windows, DOS and MacOS, run the basic functions of a computer. Unix powers many commercial and Internet computers. Unix tells time differently, which means that it does not have a year 2000 problem. Unfortunately, it does have a Sept. 8, 2001, problem. In Unix language, that date is represented by the number 999,999,999 -- the same number that some Unix applications use to denote the end of a file. Circa 2025: The U.S. telephone number problem By the year 2025 or so, the United States will simply run out of available seven-digit telephone numbers and area codes. Telephone companies will have to add digits or revamp the numbering system. That, in turn, will force software programmers to overhaul every piece of software that uses phone numbers, plus all databases and archives that store phone numbers. Jan. 19, 2038: The other Unix problem The Unix operating system tells time by counting the number of seconds elapsed since Jan. 1, 1970. But like your odometer, there are only so many places on its counter. At seven seconds past 3:14 a.m. on Jan. 19, 2038, the counters on every Unix computer in the world will be full and will roll over to "0." Many computers will assume it's either Jan. 1, 1970, all over again (who wants to relive the '70s?) or that it's the end of the world (which may be a better alternative than the preceding). Circa 2050 to 2075: The Social Security number problem By 2075, the United States will have exhausted the 1 billion unique Social Security numbers possible under its nine-digit numbering system. Year 2000 expert Capers Jones suggests that the nation must be prepared by 2050 to expand or replace the many software applications that depend on those numbers. Copyright 1998 Newhouse News Service. All rights reserved.
participants (2)
-
Martin Smoot
-
Scott Harrington