I'm respectfully requesting a two timestamp scalar argument version of TIME function. If two args are given to TIME it would return the real span of time between the two scalar time values without adding an extra day to ETAs, otherwise, if given only one argument, it keeps doing what it is doing for all the old code out there that is already coded to treat the TIMESPAN struct delivered by TIME in its peculiar non-timespan non-timestamp way when the scalar arg is an ETA