https://bugs.documentfoundation.org/show_bug.cgi?id=55960
--- Comment #34 from Melove Gupta <[email protected]> --- Submitted a patch: https://gerrit.libreoffice.org/c/core/+/201071 While investigating tdf#55960, I found that the floating-point artifacts during autofill appear to originate from the linear series calculation in "ScTable::FillSeries". I implemented a potential fix by canonicalizing the computed value using "rtl::math::approxValue" before storing it, which removes artifacts like "2.22e-16" appearing near zero when filling sequences such as "-2, -1.95, ...". I've submitted a patch but I wanted to confirm whether this is the preferred approach for handling floating-point noise in FillSeries, or if there is a better location or mechanism in the Calc codebase where such canonicalization should occur. I also added a regression unit test to ensure the issue does not reappear. -- You are receiving this mail because: You are the assignee for the bug.
