We have two varieties of timestamps in test run logs, one set which comes from outside the machine running the test, which are reasonably reliable, and another which comes from the local machine's time. For WinXP test runs, the reliable ones say that a run started at 10:00 and ended at 11:00, and the local machine time ones for that same run will say that the tests themselves started at 09:02 and ended at 09:58. Are we running them in Alaska Daylight Time, or in Pacific time with the checkbox to observe Daylight Saving time unchecked?

Philor: what is the impact of this? Just so I know where to put this priority wise.

I believe this is the reason I had to back out the patch from bug 837961. There are a number of intermittent failures in timestamp-related things that are WinXP-only which I suspect might be the result of this offset from reality plus a little randomness to the clocks overcoming the way we fudge date comparisons to get around little clock fluctuations. I have absolutely no idea what will happen around either the actual Daylight->Standard switch date in November, or, if we've also managed to get ourselves back to the version of WinXP which thinks Daylight time ends in October instead of November, what will happen on the fourth Sunday in October, which would be the 23rd. So, right now frustration for one developer and possibly several invalid intermittent-failure bugs; possibly at 2am or 3am on October 23rd or November 6th or the other really fun time, either 7 or 8 hours before one of those times, permaorange test failures across all trees.

Is it possible that XP's idea of timezones and daylight savings time predates the DST changes that went into effect in 2007? https://en.wikipedia.org/wiki/Daylight_saving_time_in_the_United_States#Second_extension_.282005.29

It's certainly possible we reverted back to the state before it knew, since we added that hotfix individually, but that doesn't explain this since that would only mean that they would change on October 23rd instead of November 6th.

There was a GPO in place that should have corrected the timezone . I am trying figure out why it wasn't firing.

A corrective GPO is going out today. The original GPO did not apply due to a hard coded password.