Leonard Mada wrote:
I noticed that Calc calculates the SUM in the same order as that of the
data cells. However I was reading somewhere (though do not remember
anymore where) that it is best to calculate the sum by:
1. ordering the numbers from smallest to largest (as absolute numbers,
i.e. without sign)
2. adding them up starting with the smallest (smallest absolute number,
i.e. 2 and not -90000 in this two number example)
If someone tried such a change, and measured the accuracy versus the
performance hit, we could decide if we want to change it.
NOTE
====
A testcase where the final result is 0 is tricky, because I believe Calc
rounds many things to 0, i.e. results that should give 0 give actually 0
(not necessarily because of brilliant calculations but because of
rounding), and results close to 0 are rounded, too, and give 0 (though
they shouldn't).
Yes, results "very close" to zero are made zero, so comparisons work
(try to explain to a user that 0.1+0.2-0.3 isn't equal to 0). See
sal/inc/rtl/math.hxx for the details.
Niklas
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
