https://bugs.documentfoundation.org/show_bug.cgi?id=165873
BogdanB <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- Resolution|WORKSFORME |FIXED CC| |[email protected] --- Comment #8 from BogdanB <[email protected]> --- Fixed by commit c6d9c76be68a8eea65d4c0c236006b7ee9ec8355 Author: Mike Kaganski <[email protected]> Date: Wed Dec 10 21:13:55 2025 +0500 tdf#169882: revert commit c799de145f7e289f31e3669646e5bd12814e6c5e (tdf#138518 sw: layout: avoid moving flys forward prematurely, 2021-04-22). That change made sure that if a fly tries to move its anchor forward, and there are more fly frames on this page anchored after this fly's anchor, then the anchor was not added to "moved forward frames" list that is used to prevent frames from moving backward. (Note that the problem is similar to the problem fixed by commit 02a1edc64bde14401899d0f026d49dbf125e3ffc - see its commit message for details.) Bugdoc for tdf#169882 has a table with a row having two cells, each with an image; the images require the row to move forward. For the first cell, layout moved the image forward (which caused the whole row to move), but didn't mark its anchor, because there was another fly "below this". Then the text succeeded moving back, creating a layout loop. Before the fix of tdf#166871 (which caused the regression addressed here), the optimization allowed the layout to skip the problem (but it was actually possible to trigger it in a different way). It could be possible to modify the code to special-case table rows. The code in SwObjectFormatterTextFrame::CheckMovedFwdCondition contained a similar condition for "on same page" case. But it seems that the whole workaround is now not needed. Reverting it keeps bugdoc from tdf#138518 fixed, so avoid the fragile workaround. This change also reverts commit d9e38d5830ac278d7180b96fb6cefe3ee8ef6d76 (tdf#135220 sw: fix layout after SwUndoDelete), which keeps its steps behave correctly. Change-Id: I299e8f78b83c1d73402e548ea7f6cce1a2e5af0d Reviewed-on: https://gerrit.libreoffice.org/c/core/+/195395 Reviewed-by: Mike Kaganski <[email protected]> Tested-by: Jenkins -- You are receiving this mail because: You are the assignee for the bug.
