Re: Temporal database - no end date

Marshall wrote:
> ...
> I used to work on a system that had a character encoding
> that used 5 1/3 bits per character. That's right: five and
> one third bits for each character. 2^5.333 ~= 40.32;
> the character set consisted of 40 characters, each consuming
> about 5 1/3 bits. If you had 3 characters, that was 16 bits.
> 5 1/3 * 3 = 16. Amazingly, math with fractional numbers
> works even on quantities that can't be divided!
> ...

Yes, from an RM perspective, I think it all depends on just exactly what are the domains one wants one's relations to operate from.

I remember another system that only cared about days, not times. Arithmetic was done on not on dates, but on an arbitrary difference between dates. The reason I didn't forget wasn't that, though. The dates were stored with 7 bits to stand for the year, 4 for the month and 5 for the day of the month. There was no such thing as a date before Jan 1, 1900. That company started to panic in the 1980's thinking they would have a year 2000 problem. I thought they should have postponed the panic 'til 2018 which would still allow ten years to solve the problem.

They weren't a manufacturing company and didn't care about time of day, otherwise they might have been better advised to add that to their systems.

p Received on Sun Jan 21 2007 - 20:44:40 CET