Re: [Gnash-commit] gnash ChangeLog testsuite/actionscript.all/Numb...

[Sandro Santilli]

On Mon, Sep 17, 2007 at 02:26:13PM +0800, zou lunkai wrote:
> strk,
> 
> see tests:
> 
> var x1 = Number(268);
> var x2 = new Number(268);
> 
> Compiled with makeswf,  dumped with listswf. got:
> 
>     SWFACTION_CONSTANTPOOL
>      [000] Number
>      [001] x1
>      [002] x2
>     SWFACTION_PUSH
>      [000]   Constant: 1 "x1"
>      [001]   Integer: 268
>      [002]   Integer: 1
>      [003]   Constant: 0 "Number"
>     SWFACTION_CALLFUNCTION
>     SWFACTION_DEFINELOCAL
>     SWFACTION_PUSH
>      [000]   Constant: 2 "x2"
>      [001]   Integer: 268
>      [002]   Integer: 1
>      [003]   Constant: 0 "Number"
>     SWFACTION_NEWOBJECT
>     SWFACTION_DEFINELOCAL
>     SWFACTION_END
> 
> So, x1 is the result of CALLFUNCTION and x2 is the result of NEWOBJECT.
> Shouldn't  they have different types as we expect?

Yes, but in Gnash this only works for user-defined constructors, not for 
built-ins
which are instead expected to return new objects (design failure).
There are TODOs here and there warning about this... in particular talking about
having built-ins and user-defined funx behave the same

--strk;