Andre Poenitz wrote:
> I don't think so.
>
> end_par is indeed the difference to the _end_, not an offset to the
> begin. This way, both values stay correct even if something is deleted
> or inserted in the middle.
>
> [I am not sure we still need this, though].
In fact I don't understand why it would be needed.
But in recordUndo we have:
stack.push(Undo(kind, textnum, pit.index(),
first_par, end_par, text->cursor.par(), text->cursor.pos()));
lyxerr << "undo record: " << stack.top() << std::endl;
// record the relevant paragraphs
ParagraphList & undo_pars = stack.top().pars;
ParagraphList & plist = text->ownerParagraphs();
ParagraphList::iterator first = plist.begin();
advance(first, first_par);
ParagraphList::iterator last = plist.begin();
advance(last, last_par);
for (ParagraphList::iterator it = first; it != last; ++it) {
undo_pars.push_back(*it);
undo_pars.back().id(it->id());
}
undo_pars.push_back(*last);
undo_pars.back().id(last->id());
If I read this correctly, the paragraphs recorded are from first_par to
last_par (inclusive).
Alfredo
