From what I knew and what I just re-read, the datetime field in a SQL server table has the same structure as in VFP - 4 byte integer to represent the date, 4 bytes integer for the milliseconds. The difference is in VFP's use of Julian day number where TSQL uses 1900-01-01 as day zero. Both have the same limitation - the values go between First Gregorian Date, to the Y10K Bug Day (9999-12-31). Both can store values outside of the range, but there's no code to display them, and the logic wouldn't really work before the FGD (which wasn't at the same time everywhere, so there were not just timezones, there were datezones).

My question, if anyone knows - why does SqlColumns() say that buffer_length=16 for datetime fields, if they occupy 8 bytes?