https://bugs.freedesktop.org/show_bug.cgi?id=87386

[email protected] changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|NOTABUG                     |---

--- Comment #4 from [email protected] ---
Based on the criteria for determining the importance of a bug, crashes and
systems slowing to a crawl are the worst. From the point-of-view of
engineering, medicine, logistic, finance, etc. there is something worse: wrong
answers that cause losses. To anyone who doesn't understand how computers store
and calculate with real numbers (most people), the 4 equations I provided
should give identical results. In LibreOffice, they don't.

Either there is a bug in how LibreOffice handles decimal calculations
internally (see bug 87506), or there is a bug or omission in how the DATE()
function in LibreOffice handles the conversion of real numbers to integers (or
both). The DATE() function shouldn't expect the user to make sure that how the
month is calculated always returns an integer for the month, because dates are
stored as real numbers, not integers (e.g., December 19, 2014 at 1:16 PM =
41992.5526594578). The month should not change because if an inaccuracy of
0.00000864 seconds (1E-010 days).

1 and 0 are integers, so all digital calculations are integer calculations.
That means some real numbers cannot be calculated digitally with 100% accuracy.
Being off by 1 in the 15th (or greater) decimal place is normally not a
problem, except where INT and MOD type functions can turn a difference of
0.000000000000001 into a difference of 1, and can cause logical values that are
true for practical purposes to be false. Most users are not aware of the
details of digital computations and their limitations, will not know how to
adjust their formulas accordingly, and will expect equivalent formulas to
result in essentially identical answers. Therefore, digital calculations must
account for this problem. In Excel and in most of LibreOffice Calc, they do;
but in this case in LibreOffice, they don't.

 In fact, if the 4 equations I provided to reproduce the problem are entered in
Excel, all 4 provide the same, correct result. The first two give correct
results in LibreOffice because mulitplying by 100 eliminates all significant
decimal digits.

-- 
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Libreoffice-bugs mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/libreoffice-bugs

Reply via email to