On Wed, Feb 04, 2004 at 10:21:00AM +0100, Alfredo Braunstein wrote: > >> Otherwise, we should ascertain that we have coordinates before (i.e. an > >> update has been called) before calling LyXText::edit(x,y)... ugly. > > > > Or put some 'sensible' value there, i.e. LCursor::getPos() or similae > > Put a sensible value where? The problem is that LyXText::edit(x,y) need > row/column coordinates information AFAICT...
Good point... *shrug* > >> Unrelated1: triple click fails because LyXText::dispatch takes a cur > >> argument but the 95% of it uses the real cursor. Smells fishy. > > > > The mouse stuff is not IU'ed yet. As a first approximation, everybody > > should use the 'cur' argument, not 'cursor()'. > > Humm.. if we are doing such a tedious change, I propose to pass around a > PitPosPair or CursorSlice instead of a full cursor... Not sufficient. For starters, it doesn't contain the bv() member. Secondly, we sometimes copy the passed cursor to the bv().cursor(). I think there are more reasons. > >> Unrelated2: MOUSE_PRESS opens a collapsable. MOUSE_RELEASE closes it. > >> Guess the outcome. > > > > No need to guess. I've seen it yesterday evening. I was, however, happy > > not to see it crash ;-) > > me too, it's kind of funny. (you can always move the cursor out of the > button before releasing)... ;-) I noticed that, too ;-) Andre'