As has been pointed out in A billion seconds of Unix, the Unix clock rollover on Saturday, September 8th, 2001 is liable to cause some breakage in scripts that either assume that they can compare raw time values as strings, or which assume that the resulting string is 9 digits.

This isn't quite the scale of Y2K, but regardless is one of the unique circumstances that we get to experience in our lifetime. So let's keep a record.

As you run across clock-rollover induced script breakage, post a reply, and list the culprit.


In the interest of fairness, I did find a cmp that should have been a <=> in one of my personal scripts...)

Replies are listed 'Best First'.
Re: 1E9 Seconds Hall of Shame (followup)
by dws (Chancellor) on Sep 10, 2001 at 23:40 UTC
    A Slashdot thread on the topic of (time_t) 1e9 rollover bugs reports the following:
    • At least one person storing a 9 digit date in MySQL
    • A KDE 1.x KMail bug that prevents correct date display
    • A report that the KDE news reader isn't ordering articles correctly
    • A pointer to a bug report for IkonBoard that effects article sorting (for several versions of the product).
    • A bug in some versions of CVSup that mangles dates
    • A possible problem with Canon S10 software (on the PC side?) getting dates wrong
    • Date-sorting breakage in some ISP's IMAP server
    • Mention that Veritas has issued a warning that their backup software displays dates as 1970 (truncation?)
    • A report of "massive breakage" in OpenLDAP 1.0 and 2.x
    Perl is implicated in a few of these problems.

    Not quite Y2K (and probably not as spectacular as the 2038 out-of-bits rollover is liable to be), but interesting nonetheless.

[reply]

Back to Meditations