On Mon, Jul 15, 2002 at 09:08:13AM +0200, Juergen Vigna wrote:
>
> I would say because there are actions not given by a cursor movement
> which will have to move the cursor to the position. Maybe UP is not
> one of them but _before_ and _behind_ are needed so I think we should
> have all 4 directions and do not have to reinterpret lfuns outside
> again. What problem do you have with this? Heed that just FINISHED
> is not enough as result value.
For RTL I need to have a distinction between FINISHED_LEFT/RIGHT
and FINISHED_BEFORE/AFTER, so in other words, we will need to have
6 FINISHED* values.
But looking at the source, the only places where FINISHED* is returned when the
action is not LFUN_LEFT/RIGHT/UP/DOWN are:
1) in insettext.C/insettabular.C
if ((action < 0) && arg.empty())
return FINISHED;
Why is this needed ? Can action be < 0 ??
2) In formulabase.C
if (!arg.empty()) {
bv->lockedInsetStoreUndo(Undo::EDIT);
if (arg.size() == 1)
result = mathcursor->interpret(arg[0]) ? DISPATCHED : FINISHED_RIGHT;
This is used to exit from the inset when pressing space at the end.
As I said, we can only have one FINISHED value. If the action performed was
LFUN_LEFT/RIGHT/..., then it dictates the cursor movement in the parent
paragraph. Otherwise, the cursor is put after the inset.
