Re: [Gnash-dev] Problem with Drawing API

[zou lunkai]

> I think I never saw DrawingAPI producing color leakage.
> Can you produce such thing ?

Hmm, not sure what you mean.

Try the following example, you'll see the differences between Gnash
and PP.  Conceptually, we need the
'normalization' step to finish the last two lines(commented in the
example code) if user forget to do that.  And this
step is probably more complex  in  general cases and we don't have any
specs to define the "expected" behaviour.
That's why I said drawingAPI users should be responsible for their shapes.


example:

_root.createEmptyMovieClip("mc", 1);
with(mc)
{
        beginFill(0x0000ff, 100);
        lineStyle(5, 0xff00ff, 100);
        moveTo(200, 200);
        lineTo(300, 200);
        lineTo(200, 300);
        lineTo(300, 300);
        // lineTo(200, 200);  // with or without these two lines doesn't make
a difference with pp.
        // endFill();
}

--zou


On Nov 6, 2007 3:53 PM, strk <address@hidden> wrote:
> On Tue, Nov 06, 2007 at 09:51:43AM +0800, zou lunkai wrote:
> > > I had the normalization step in mind while doing the DrawingAPI interface
> > > so we have a ::finalize method in DynamicShape which should be called 
> > > before
> > > any function that needs normalization of it (rendering, hit test, bounds
> > > computation) and that uses a "dirty" flag set whenever something is drawn.
> > > Currently the function does nothing.
> >
> > For swf defined shapes, the normalization step is done by the pp
> > compiler. There should be a question for the dynamic shapes created by
> > DrawingAPI: who should take care of the  normalization?  DrawingAPI
> > interface?  renderer?  or the API users?
> >
> >  I  guess the API users  should be  responsible for not producing
> > malformed shapes. It is not a trivial work for a *decoder*  to produce
> > a satisfied normalized shape with random inputs.
>
> I think I never saw DrawingAPI producing color leakage.
> Can you produce such thing ?
>
> --strk;
>