To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=72967
------- Additional comments from [EMAIL PROTECTED] Thu Jan 25 16:48:43 -0800 2007 ------- I made the following experience in Calc 2.1, Excel 2003 and Scilab 4.1, and I found in all cases the same error: 2701.200300000000-2701.000000000000-0.200300000000 is not equal to zero, but -2.984D-14 (CPU: AMD Athlon) It may be difficult to do better in Calc than in Scilab, whose purpose is really to make scientific calculation. But it may be interesting to secure several functions whose results strongly depend on this float approximation error: - One of them is the function DATE(year, month, day) which takes the floor() of the year, month and day arguments. Take the previous example: the two first arguments taken separately return 0.2003-2.984D-14, which is <0.2003. This value*1000 corresponds to the year 2003, but in the DATE() function, it is understood for the year 2002!! My brother told me, that such an error may result into the fact, that the next world war could happen one year before our expectations!! No seriously, I controlled in Excel: the DATE() function is securized and this error does not occur. - DATE() is one method (which interested me), but IF() is also a method whose result can have serious consequences. In fact, any method converting floats into boolean or integer, or any other discrete type is subject to be affected by this approximation. Good luck, (I may spend a little bit time on this topic, if required) Jerome RG --------------------------------------------------------------------- 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]
