when we call into nsIWidget::EndResizingChildren under linux, the widget code calls a gtk thaw method. this method synchronously forces a redraw of the entire widget. there are two problems with this: 1) it draws the entire widget regardless of any children moving/resizing (or not). 2) the refresh is synchronous. it should call the queued drawing methods instead. this makes demo13 run *much* slower than it should.

the whole gtk event model wrt gecko is busted, imho. I spent many hours debuggin it last week and came to the conclusion that gtk is doing too much without a simple fallback, for example, gtk event percolate from a child window to its parent and so on until it is handled by one of them. So, motion events percolate up the widet hierarchy and there is not way to break from the loo other than iterating the gtk layout children and clipping them with the motion event coordinates and dropping the event if they match. And other such insanity. Also, gtk is doing all kinds of real event and synthetic signal processing. Finally, the gtk layout widget has its own idea of how to layout children. A lot of this functionality is general purpose and fine for simple applications, but it makes it hard to optimize gecko for gtk. Im looking at these issues right now. marking assigned m5

i wont get to this till m6 or later. marking m6.

marking m9

Moving all Widget Set bugs, past and present, to new HTML Form Controls component per request from karnaze. Widget Set component will be retired shortly.

moving to XP Toolkit/Widgets since these are not HTML form controls

mass-moving all m12 bugs to m13

we no longer use gtklayout's so this is fixed.

marking verified