Re: [Gnash-commit] gnash ChangeLog server/as_value.cpp server/as_v...

[Sandro Santilli]

On Tue, Nov 27, 2007 at 08:53:05AM +0800, zou lunkai wrote:
> +       /// @param allowUnloaded
> +       ///     If true an unloaded sprite is still returned as such,
> +       ///     rather then attempted to be resolved as a soft-reference.
> +       ///     Main use for this is during paths resolution, to avoid
> +       ///     infinite loops. See bug #21647.
> +       ///
> +       sprite_instance* to_sprite(bool allowUnloaded=false) const;
> 
> In which scenario we are not allowed to access an unloaded sprites?
> Shouldn't  to_sprite() always allow unloaded characters?

I tried to produce a testcase with no success yesterday.
Yes, the wording is confused, we're always allowed to access
an unloaded sprite, the difference here is wheter to attempt
to rebind a soft-reference to an unloaded sprite or not.

The signature could be renamed to take a bool skipRebinding=false

The problem was a _level0.ch.ch path that while resolving the
first 'ch' element reached the as_value::to_object() point
(ch was found as a proper property but not as a child).
to_object() in turn called to_sprite as it was a soft-reference
which in turn tried to evaluate the target which got to the start
(maybe a soft-reference to self, with self unloaded would do, probably
inside an onUnload handler; will try again today).

--strk;