> Numeric arguments are no longer interpreted as Julian Date numbers. No. REAL values are interpreted as Julian Dates and INT values are interpreted as unix timestamps. Do you know of an application that stores an unix timestamp as a REAL or a Julian date as an INT? I do not. So, I'm wondering if the change will have any impact at all. > all pre-existing code that depends on the current behaviour \[must be\] modified The only applications that would need to be modified are those that store Julian Dates as INT, or that store Unix timestamps from the first three months of 1970 as REAL. Do you know of any such applications? I'm guessing that the number of such applications is zero, at least to a good approximation. Do you know of any counter-examples?