Hi, Le 01/07/2014 03:59, Joel Madero a écrit : [...] > This is indeed well known and Kohei has explained why it happens (quite > a few times). The logic is beyond my comprehension but I trust his > judgment 110% and he has said it is really unfixable. I believe we have > closed at least 5 similar bugs in the past year.
Yes, I share the judgment of Kohei and Markus that the rounding errors problem can't be solved for all computations without losing performance. Infinite precision is too expensive. Current behavior is not problematic if we keep rounding errors small. It is not always the case and we should try to avoid the cases where rounding errors gives final errors that are not small. That is the challenge of Numerical Analysis, finding correct results with false computations. ;-) In comment #2 of https://bugs.freedesktop.org/show_bug.cgi?id=80732 I showed that it is possible to make better computations of MOD() and QUOTIENT() functions. At least in the cases reported here. Best regards. JBF -- Seuls des formats ouverts peuvent assurer la pérennité de vos documents. _______________________________________________ List Name: Libreoffice-qa mailing list Mail address: Libreoffice-qa@lists.freedesktop.org Change settings: http://lists.freedesktop.org/mailman/listinfo/libreoffice-qa Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/ Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette List archive: http://lists.freedesktop.org/archives/libreoffice-qa/