70762 2011-10-24 14:32:24 -0700 Incorrect repaint during layout in flipped writing modes 2025-09-08 09:38:35 -0700 1 1 1 Unclassified WebKit Layout and Rendering 528+ (Nightly build) Unspecified Unspecified NEW https://bugs.webkit.org/show_bug.cgi?id=279611 BrowserCompat P2 Normal --- 1 mitz webkit-unassigned ahmad.saleem792 hyatt ntim simon.fraser zalan oldest_to_newest 489646 0 112248 mitz 2011-10-24 14:32:24 -0700 Created attachment 112248 Test case In the test case, edit either line and notice that it does not repaint correctly. The reason is that RenderBox::computeRectForRepaint() calls flipForWritingMode() during layout, when the block’s height is still 0. 1942291 1 ahmad.saleem792 2023-03-18 06:26:43 -0700 I am able to reproduce this bug in Safari Technology Preview 165 using attached test case and Safari show 'ipsum.Lorem' in two lines with newlines in between and at top. While Chrome Canary 113 and Firefox Nightly 113 match each other. Thanks! 1942301 2 zalan 2023-03-18 09:06:28 -0700 This is about horizontal-bt writing mode. Neither FF nor Chrome support this (non-standard) mode. 2060071 3 zalan 2024-09-13 11:54:59 -0700 While the test case has horizontal-bt, as the title says this is about flipped writing modes in block direction mostly impacting vertical-rl. 2060081 4 zalan 2024-09-13 12:10:55 -0700 The issue here is with flipped block direction an inflow child's final position (in absolute repaint coords) is unknown until its containing block finished sizing (unless containing block is the paint container). Our current repaint-after-layout logic runs descendants-first which is too early for such flipped writing modes. In many cases though this is covered up by repaints issued _before_ layout initiated by style or content change. 2141610 5 ntim 2025-09-08 07:40:15 -0700 There's was a workaround added here for this issue: https://github.com/WebKit/WebKit/commit/8896aad97fa3f274318189a550b3e0abb25780ba#diff-905e44c8aef545da6266ae6086148ba389183bff128d9a4eb2bfd9e9987a02e8 2141627 6 mitz 2025-09-08 09:38:35 -0700 Just for clarity, the workaround is specifically for some form controls rather than all possible manifestations of this issue, right? 112248 2011-10-24 14:32:24 -0700 2011-10-24 14:32:24 -0700 Test case repaint-during-layout-in-flipped-block.html text/html 141 mitz PGRpdiBzdHlsZT0iLXdlYmtpdC13cml0aW5nLW1vZGU6IGhvcml6b250YWwtYnQ7IC13ZWJraXQt dXNlci1tb2RpZnk6IHJlYWQtd3JpdGU7Ij4KICAgIDxwPgogICAgICAgIExvcmVtPGJyPgogICAg ICAgIGlwc3VtLgogICAgPC9wPgo8L2Rpdj4K