"G Ashton" <[email protected]> wrote:
|to test inputs to be sure they are in the domain of the function. I have
|found that many
|published algorithms fail to state the earliest and latest date for which
|they work. Finding out
|will require much more than 10 seconds.
The function that has been stolen via third hand from
Communications of the ACM, Vol 6, No 8. is documented as working
for the period in between, and including, 01-01-01 and
65535-12-31, except that non-existing dates are unknown and leap
year calculations do not work correctly until the final
introduction of the gregorian date, 1582-10-15.
I hope that is still correct after so many copies.
--steffen
--- Begin Message ---
Clive D. W. Feather wrote, with respect to conversion between JDN and
Gregorian calendar date,
> >So in order to calculate the
>> actual date where the drift adjustment occurs you have to face a very
> >elaborate conversion.
>No, you need to use a library that's already been written to do the job.
>Takes 10 seconds or so.
To make people comfortable with your work, you need to state the earliest
and latest date
For which your code works correctly. If your organization is strict, you
might even be required
to test inputs to be sure they are in the domain of the function. I have
found that many
published algorithms fail to state the earliest and latest date for which
they work. Finding out
will require much more than 10 seconds.
Of course the code has to be tested. A list of known-good conversions should
be easy
to find for contemporary dates. Although not needed for the purpose of leap
seconds,
does anyone know a known-good conversion list that includes ancient dates
and dates
far in the future?
Gerard Ashton
_______________________________________________
LEAPSECS mailing list
[email protected]
https://pairlist6.pair.net/mailman/listinfo/leapsecs
--- End Message ---
_______________________________________________
LEAPSECS mailing list
[email protected]
https://pairlist6.pair.net/mailman/listinfo/leapsecs
