[evlatests] timing tic / Freaky Fridays.

[Ken Sowinski]

On Sun, 1 Mar 2009, Ken Sowinski wrote:

> On Sat, 28 Feb 2009, Vivek Dhawan wrote:
>
>> The current theory is that the weekly EOP updates are implicated.
>
> The implication is only circumstantial.  EOP parameters are fetched
> from USNO on Friday and the time glitch occurs Friday night.  It is
> hard to see why this should have started on February 14 when the last
> change to relevant software was much earlier.

Some of the history.  Barry reported that the change for the leap
second  introduced this.  That change was commited Dec 30 and used
at the turn of the new year.  The standard executor was used after
the New Year recess.  The update which included this change was on
Wednesday, Jan 14.  Thus, except for the one USNO server failure,
we have been vulnerable to a time error every Friday since then.