Jean-Marc Lasgouttes wrote:
> Juergen> Indeed, but it does not only care for dEPM, but also if the
> Juergen> cursor is valid. I wonder, though, if the explicit set of
> Juergen> bvcursor is not redundand now.
>
> It may be interesting to experiment.
I commented those lines out and didn't see any bad effects at first glance. It
looks like this:
if (cmd.button() == mouse_button::button1
|| (cmd.button() == mouse_button::button3
&& (&bvcur.selBegin().inset() != this ||
!tablemode(bvcur)))) {
cur.selection() = false;
setCursorFromCoordinates(cur, cmd.x, cmd.y);
// cares for dePM
cur.bv().mouseSetCursor(cur);
//cur.resetAnchor();
//bvcur = cur;
break;
}
while BufferView::mouseSetCursor does:
void BufferView::mouseSetCursor(LCursor & cur)
{
BOOST_ASSERT(&cur.bv() == this);
// Has the cursor just left the inset?
bool badcursor = false;
if (&cursor().inset() != &cur.inset())
badcursor = cursor().inset().notifyCursorLeaves(cursor());
// do the dEPM magic if needed
// FIXME: move this to InsetText::notifyCursorLeaves?
if (!badcursor && cursor().inTexted())
cursor().text()->deleteEmptyParagraphMechanism(cur, cursor());
cursor() = cur;
cursor().clearSelection();
cursor().setTargetX();
finishUndo();
}
so cur.bv().cursor() is set in mouseSetCursor. Is this enough? What could
happen?
Meanwhile, I have committed the previous patch.
Jürgen
