To comment on the following update, log in, then open the issue:
http://www.openoffice.org/issues/show_bug.cgi?id=101316





------- Additional comments from [email protected] Sun Apr 26 21:20:38 
+0000 2009 -------
The error comes from cancellation with total loss of significance in value-mean
in some cases. It seems to be the same kind of error as the known "3*1/3 != 1"
problem.

I see the error only, when using identical values in the data series, which
seems to be a very constructed example. But the results are totally wrong in
those cases. I'll attach test cases, where you see the error examples with
colored background. You can notice, that the values are correct, if the
denominator is 4, which can be represented exactly in binary code.
I looked at Gnumeric and Excel too, neither of them correct this kind of 
errors. 

You can correct the error, if you use approxSub instead of a simple minus. I
have tried to find out, whether this change will decrease accuracy in other
cases. I don't find any examples, which would have a decreased accuracy, but
perhaps you have got some critical test case?



---------------------------------------------------------------------
Please do not reply to this automatically generated notification from
Issue Tracker. Please log onto the website and enter your comments.
http://qa.openoffice.org/issue_handling/project_issues.html#notification

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to