The next looming, slightly less serious, 2 digit year problem

Roughly 15 years ago there was an intense crisis as people realised that the results of only storing the final two digits of a year were about to become quite ambiguous.1

A variation of this ambiguity still exists; date-parsing functions such as strptime have to deal with it all the time. The standards2 pick a pivot value—69—and everything on or after that is interpreted as 19XX, everything before as 20XX.

For example:

#include <stdio.h>
#include <time.h>

int main(void)
{
    struct tm result;
    char buf[255];

    if (strptime("68", "%y", &result)) {
        strftime(buf, sizeof(buf), "%Y", &result);
        puts(buf);              /* 2068 */
    }

    if (strptime("69", "%y", &result)) {
        strftime(buf, sizeof(buf), "%Y", &result);
        puts(buf);              /* 1969 */
    }

    return 0;
}

By my reckoning, this means we are heading for another crisis in about 45 years.3 I suggest you start stocking your bomb shelter with non-perishables now.


  1. It was so intense that there was almost a feeling of rage afterwards; “What do you mean, we spent all this money preventing it, and then nothing actually happened??” ↩︎

  2. X/Open Common Application Environment and POSIX ↩︎

  3. Unless of course you believe the earlier-looming 2038 problem spells doom first. ↩︎


187 Words

2014-05-28 00:00 +0000

comments powered by Disqus