Created attachment 121218 [details] Wrong zero calculation Simple formula (=C26+A27-B27) calculates "5,6843418860808E-14" instead of expected "0". System: Linux Mint 17 (Linux 3.13.0-24-generic (x86_64)) Version: 5.0.3.2 Same behavior using LO 5.0.2.2 with Windows XP (32). See cell C26 in the attachement. Changing cell A10 to "181,000005" and B26 to "178,500005" results in correct value "0" in cell C26. With values "181,000006" and "178,500006" the error occurs.
It's a rounding problem with floating point numbers. Annoying maybe, but not a bug I guess. When set options > lo calc > calculate > 'limit decimals ....' on and 'decimal places' to 2, the problem is gone.
I think the issue usually occurs with numbers that are periodic in binary base, so doing operations on them results on the precision problems. If you set up 15 decimal places for the all cell's format, you can see that C27 is different than B27 on the last decimals. The precision in calc is fifteen numbers plus sign, and this is a hardware limitation. https://bugs.documentfoundation.org/show_bug.cgi?id=50299 take special attention to comments 4 and 6 of Kohei, one of the most expert developers with calc. Combining the MM mentioned option with 12 decimal places, with Menu/Tools/LibreOffice calc/Calculate - Precision as Shown seems to work fine in this case. What usually for accounting purposes it's enough precision. Please if you are not agree with resolution, reopen it.