Another windowing questiongreenspun.com : LUSENET : TB2K spinoff uncensored : One Thread
If a system only handles dates within a short timespan, wouldn't an unremediated system fix itself and stop making date errors once all relevant data were from 2000?
Could some systems have been "fixed" this way, by manually correcting errors for a short time until the "error window" closed normally? How many?
-- Flint (firstname.lastname@example.org), March 14, 2000
Could be - if the system didn't update any databases in the "short period" when it was wrong, and if later runs didn't depend on the accuracy of earlier runs.
But I don't this this is very typical. I didn't see any.
-- kermit (email@example.com), March 14, 2000.
Flint, One of our control systems was not compliant and we were unable to get an update from the manufacturer in time for the rollover. To get by we set the clock back on that system, and once the rollover was completed, we set it forward again to the correct time. It is now working perfectly in the year 00.
-- Malcolm Taylor (firstname.lastname@example.org), March 14, 2000.
Not entirely. Some errors are caused by virtue of having leading zeros...
Say you're using yyddd julian dates (00074 for today). If you make it a numeric field, then redefine it as alphanumeric for display purposes, some languages (Ideal, for ex.) assume you want to suppress leading zeros. The next time you refer to the numeric version of the data - burp. We had a bunch of those.
Add to that the lookback calculations which subtract 1 from 00 - though those will dwindle over the course of this year and be gone next year, if you can wait that long.
-- RC (email@example.com), March 14, 2000.