Currently the watchdog() function is logging events using REQUEST_TIME constant which is just $_SERVER['REQUEST_TIME']. This works fine in most cases but processes that take an extended period of time all get logged at the time the process began, and not when the event actually happened. For example when running data migration scripts that can take much longer to process all of the errors get logged at the time the process began. It seems that watchdog would be more accurate if it logged the time with time() instead.
PASSED: [[SimpleTest]]: [MySQL] 39,407 pass(es).
PASSED: [[SimpleTest]]: [MySQL] 41,106 pass(es).
PASSED: [[SimpleTest]]: [MySQL] 35,645 pass(es).
FAILED: [[SimpleTest]]: [MySQL] 35,636 pass(es), 9 fail(s), and 1 exception(s).
FAILED: [[SimpleTest]]: [MySQL] Unable to apply patch drupal-1239410-7.x.patch. Unable to apply patch. See the log in the details link for more information.