Find a way to minimize the number of save/restore calls.
Created attachment 79917 [details] Patch
Comment on attachment 79917 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=79917&action=review > Source/WebKit/qt/ChangeLog:9 > + In the first loop of renderRelativeCoords() the call of QPainter::save/restore is useless, because > + the only change is on it's context which is restored. This sounds a bit strange, I would say something like "..., because the context is saved/restored within the loop." > Source/WebKit/qt/ChangeLog:10 > + In the second loop their calling is also avoidable by using invert translation on context. s/their calling is/the calls are/ s/invert/inverse/ > Source/WebKit/qt/Api/qwebframe.cpp:408 > view->paintScrollbars(context, rect); Are you sure that FrameView::paintScrollbars() doesn't taint the GraphicsContext state?
(In reply to comment #2) > (From update of attachment 79917 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=79917&action=review > > > Source/WebKit/qt/ChangeLog:9 > > + In the first loop of renderRelativeCoords() the call of QPainter::save/restore is useless, because > > + the only change is on it's context which is restored. > > This sounds a bit strange, I would say something like "..., because the context is saved/restored within the loop." > > > Source/WebKit/qt/ChangeLog:10 > > + In the second loop their calling is also avoidable by using invert translation on context. > > s/their calling is/the calls are/ > s/invert/inverse/ Uppppdated... > > Source/WebKit/qt/Api/qwebframe.cpp:408 > > view->paintScrollbars(context, rect); > > Are you sure that FrameView::paintScrollbars() doesn't taint the GraphicsContext state? No, GC is absolutely undamaged :)
Comment on attachment 79917 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=79917&action=review >>> Source/WebKit/qt/Api/qwebframe.cpp:408 >>> view->paintScrollbars(context, rect); >> >> Are you sure that FrameView::paintScrollbars() doesn't taint the GraphicsContext state? > > No, GC is absolutely undamaged :) Awesomecake! r=me
Closing bug. This is landed in <http://trac.webkit.org/changeset/76516>
http://trac.webkit.org/changeset/76516 might have broken GTK Linux 32-bit Debug The following tests are not passing: editing/selection/extend-selection-bidi.html