Libre Office v5.0.2.2 on Windows 10. Adding a simple formula for more than 100 rows results in correct display. After saving and closing the file you reopen it. The unchanged formula shows a wrong calculating. If you want to see the wright result you delete one cell containig a formula at row 2 and copy the upper or lower cell to it. Voilà the correct result appears. The solution of "A (Andy)" in Bug 91865 https://bugs.documentfoundation.org/show_bug.cgi?id=91865 is not resolving the bug. See attached PDF file with step by step screenshots and two calc files: - "bug v5.0.2.2.ods" -> bug file with more than 100 rows. - "bug v5.0.2.2 - OK.ods" -> no bug, less than 100 rows. Regards Silvio
Created attachment 119000 [details] step by step screenshots
Created attachment 119001 [details] bug file with more than 100 rows
Created attachment 119002 [details] no bug, less than 100 rows
Duplicate of bug 94271 ?
(In reply to MM from comment #4) > Duplicate of bug 94271 ? No, it's not a duplicate. I have always the same wrong result and not always other wrong results. See my formula: it's a very simple addition of "cell n + 1".
Not reproduced from scratch and not confirmed with attachment 119001 [details]. I don't get the phenomenon of 2,2,2,2... in column A etc. Win 7 Pro 64-bit, Version: 5.0.2.2 (x64) Build ID: 37b43f919e4de5eeaca9b9755ed688758a8251fe Locale: fi-FI (fi_FI)
I am also seeing this bug. On my machine I can reproduce this every time with the steps below on this very small file (attached -- testcase.ods) Steps to reproduce --------------------- 1) Open file testcase.ods 2) Change cell C35 (highlighted in yellow) to any other value; try zero. 3) All calculations in red rows in column E turn to zero. 4) Those cells in column E should not be zero - they are simple additions of two nonzero cells in the sheet. My system ----------- Libreoffice 5.0.2.2, installed directly from LibreOffice website link Windows 7, 64bit -- all security patches applied
Created attachment 119462 [details] Small example sheet which exhibits miscalculation
(In reply to Billy Charlton from comment #7) > Steps to reproduce > --------------------- > 1) Open file testcase.ods > 2) Change cell C35 (highlighted in yellow) to any other value; try zero. > 3) All calculations in red rows in column E turn to zero. > 4) Those cells in column E should not be zero - they are simple additions of > two nonzero cells in the sheet. I tried it, but only E35 turned zero. Win 7 Pro 64-bit, Version: 5.0.2.2 (x64) Build ID: 37b43f919e4de5eeaca9b9755ed688758a8251fe Locale: fi-FI (fi_FI)
Huh. I tried the same file on a different Windows 7 machine and cannot reproduce. No idea what's different. I'll keep experimenting and report back if I can get it to repro on this 2nd Windows machine.
Please try with a new profile: https://wiki.documentfoundation.org/UserProfile If it's only happening on a single machine then it's likely a corrupt profile.
File opens fine for me: Ubuntu 15.04 x64 LibreOffice 5.0.2.2 If it's an issue it's either the profile corrupted or Windows only. I don't see how this could be widespread as we almost definitely have unit tests for this kind of stuff.
Just an idea : Look if OpenCl is enabled. see https://bugs.documentfoundation.org/show_bug.cgi?id=94924
YES! Disabling OpenCL and then recalculating completely eliminates the problem. So this does appear to be a dupe of https://bugs.documentfoundation.org//show_bug.cgi?id=94924 Note, the machine exhibiting this problem with OpenCL has an Nvidia GTX 970 with the latest drivers. The second test machine I tried has a basic motherboard graphic (I think Intel-based).
*** This bug has been marked as a duplicate of bug 94924 ***