It would break timezone reporting for historical Moscow timestamps, ie they'd be printed as GMT-3 whether or not that was applicable at the time. Don't know whether you care about that.
The column storing datetime is of type "timestamp" and timezone is not being stored. So should not be an issue for this particular use-case. For "timestamp with time zone" that was definitely an issue. By the way won't that be issue anyways once the Moscow timezone is corrected in upgraded postgres version?
