25269 2009-04-17 10:42:01 -0700 REGRESSION: fast/repaint/fixed.html paints too much 2011-07-15 16:52:29 -0700 1 1 1 Unclassified WebKit CSS 528+ (Nightly build) Mac OS X 10.5 NEW InRadar P2 Normal --- 25602 1 eric webkit-unassigned ddkilzer hyatt jchaffraix mitz simon.fraser oldest_to_newest 117941 0 eric 2009-04-17 10:42:01 -0700 run-webkit-tests -p fast/repaint/fixed.html shows the failure. This is a regression, so technically p1, but I feel bad marking it such. ;) 117943 1 mitz 2009-04-17 10:44:52 -0700 <rdar://problem/6802446> 118162 2 simon.fraser 2009-04-20 14:38:24 -0700 The repaint that does not cover the green square is coming out of RenderLayer::updateLayerPositions(), painting m_repaintRect which has the pre-scroll location, and happens via the call to layout() from externalRepresentation(). Repaints of fixed position elements when the page has not just scrolled are not too large, so we're not over-painting in the common case. 438060 3 jchaffraix 2011-07-15 16:52:29 -0700 I looked at the different outputs of fast/repaint/fixed.html and they look fine on Chromium and Mac. Gtk has a bad image checked-in as it does not seem to implement createBitmapContextFromWebView properly.