Re: [Gnash-commit] gnash ChangeLog server/button_character_instanc...

[Udo Giacomozzi]

Monday, March 12, 2007, 8:45:52 PM, I wrote:
UG> Log message:
UG>         implement instance names for buttons and text fields

Again some additional notes:

- this patch makes sure that buttons and text fields get their
  implicit instance names (previously only sprites had that)
  
- the patch basically works but the instance names are not really
  correct because button childs get their name immediately whereas
  sprites get it only when added to the display list. This means a
  unnamed sprite inside a unnamed button gets a name like
    _level0.instance2.instance1
  whereas is should be;:
    _level0.instance1.instance2
  At least it does get a name now...

- This problem could be solved probably with a special function
  character::assignInstanceNames() that recursively walks the object
  tree and assigns names to instances that have no name but should
  have one. This function should be called *after* setting up a new
  display list and *before* executing action queues. But maybe this
  makes no sense at all...

- Again, when having a sprite inside a button, this sprite can't
  currently access the button through "_parent". Don't know where the
  problem is (spent a hour digging) but
    trace(this._parent)
  simply returns nothing (not even "undefined").

- I also noticed that with no button at all "this._parent" returns
  "_level0." (notice the extra dot).

I commited what I have in the meanwhile because there should be no
regression anyway and I can't continue working on Gnash until the end
of the week...

Greetings,
Udo