text in runs and changed the text transformer code to do the buffering
instead. It was already copying the transformed text into its internal
buffer anyway, so this saves the extra copy
git-svn-id: svn://10.0.0.236/trunk@65191 18797224-902f-48f8-a5cc-f745e15eee43
- We no longer display alternate content for broken images inside image
- We only display the image loading icon during the initial load of an image not if its src attribute is changed via the DOM.
- In AttributeChanged(), se don't call ContentChanged() on the document to reflow the image frame but call the parent frame's ReflowDirtyChild() method instead.
git-svn-id: svn://10.0.0.236/trunk@64808 18797224-902f-48f8-a5cc-f745e15eee43
that measures chunks or words. Because that function is only implemented
on Win32 the code is #ifdef _WIN32
git-svn-id: svn://10.0.0.236/trunk@64698 18797224-902f-48f8-a5cc-f745e15eee43
fixes layout of pages where a single line is impacted by 2 or more floaters
r=troy
git-svn-id: svn://10.0.0.236/trunk@63783 18797224-902f-48f8-a5cc-f745e15eee43
all I did was initialize the members of a struct in the constructor
r=troy
git-svn-id: svn://10.0.0.236/trunk@63496 18797224-902f-48f8-a5cc-f745e15eee43
In the following files, replacing disp->mVisible with disp->IsVisible() is a BUG FIX: nsBlockFrame, nsBulletFrame, nsHRFrame, nsHTMLContainerFrame, nsImageFrame, nsTextFrame.
In all the other files, replacing disp->mVisible with disp->IsVisible() or disp->IsVisibleOrCollapsed() is merely a SYNTACTICAL CHANGE. The respective owners will later review the use of IsVisibleOrCollapsed() and in some cases replace it with IsVisible(). See bug 21701.
git-svn-id: svn://10.0.0.236/trunk@63284 18797224-902f-48f8-a5cc-f745e15eee43
