https://bugs.freedesktop.org/show_bug.cgi?id=80732
--- Comment #2 from Jean-Baptiste Faure <jbf.fa...@sud-ouest.org> --- Created attachment 102047 --> https://bugs.freedesktop.org/attachment.cgi?id=102047&action=edit augmented version of test spreadsheet I agree that the current behavior is due to a limitation of the floating point calculations on the current computer technology. That said it is possible to workaround the problem in the case of MOD() function by changing its algorithm. For example by computing MOD(X;Y) = X-ROUND(X/Y)*Y Using INT(X/Y) instead of ROUND(X/Y) gives the same problem as the current algorithm because in some cases the quotient X/Y is inferior by one unit to the correct value due to the fact INT(A) make a truncation to the greatest integer less than A instead of a rounding to the nearest integer as ROUND() does. There is the same problem with the QUOTIENT() function. Computing it by QUOTIENT(X;Y) = ROUND(X/Y) gives the correct results when computing it by INT(X/Y) gives the same errors. Please have a look at the second sheet of the new version of the attachment. In column F you can check that the computed values are always near zero: select the column and verify that the sum in the status bar is small. Best regards. JBF -- You are receiving this mail because: You are the assignee for the bug.
_______________________________________________ Libreoffice-bugs mailing list Libreoffice-bugs@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice-bugs
