Created attachment 122986 [details] The file that triggers the bug How to reproduce: Open the attached file, Sheet "Autoregressive Process AR(2)". Change C5 to "0,999". All the values in the C column should be around 200, yet in LibreOffice 5.1+ starting from C14 it's something small and unrelated. If one copies the exact formula from C14 to some other place, it will calculate a correct result different. Tested in Archlinux LO 5.0.5 - works as expected Archlinux LO 5.1.0 - wrong Windows 7 LO 5.1.1.1 - wrong
I can confirm with Version: 5.2.0.0.alpha0+ Build ID: 91a7580e03d5b47c6e2513afce85ddee45e730b6 CPU Threads: 4; OS Version: Linux 4.2; UI Render: default; TinderBox: Linux-rpm_deb-x86_64@70-TDF, Branch:master, Time: 2016-02-04_00:17:43 works OK in 5.0.5.2, regreesion
I'm seeing slight incremental variations around 200 (overall variation +- 5 across the range c12 - c2037) There are no big jumps / changes of the kind described - I figure this is normal behaviour so - Unable to reproduce. LO 5.1.1.2 x64 Win 7 Build ID: fe4d9e69c82c6ee6db3c27cd5e2d47558afa80ac
Confirm, newest daily build Win x64 has the issue fixed. Build ID: 95b10e373d117f3ed3e53987aa40a65d1ca6d1a0 And the latest Windows x86 5.1.1.1 pre-release from the official website works good. Build ID: c43cb650e9c145b181321ea547d38296db70f36e I consider the issue fixed and wait for 5.1.1 to become stable and get to Arch. Thank you!
Correct status = WorksForMe (Fixed is used when a code commit is involved)
Set version from description. Best regards. JBF