On Tue, Nov 06, 2007 at 01:20:20AM +0100, Udo Giacomozzi wrote:
> Hello strk,
>
> Tuesday, November 6, 2007, 12:20:26 AM, you wrote:
> s> It seems you got the "normalization" thing by now :)
>
> I was thinking at normal shapes (coming from the SWF) which don't have
> this problem :)
We can produce all weird cases with Ming, ready to give them a try ?
Also, the testcase you attached was malformed too, I guess your new
implementation fails in that case.
> s> Currently the function does nothing.
>
> Do you have any idea how such a normalization would work? I guess it's
> not that easy since you need to split edges (the "X" case) and reorder
> them, including choosing the correct fill side. Uh oh... :(
We need to produce testcases to expose bugs.
The "old" point_test failed on the curve, but worked for
both the any-clock-wise-with-left-fill and the X case:
createEmptyMovieClip("inv1", 1);
with (inv1)
{
moveTo(10, 10);
beginFill(0xFF0000);
lineTo(20, 10);
lineTo(10, 20);
lineTo(20, 20);
endFill(); // should close to 10,10
}
inv1._xscale = inv1._yscale = 400;
inv1.onRollOver = function() {};
I just tested with revision 1.44 of shape_character_def, and the above
works fine (while it fails with your new implementation).
--strk;
_______________________________________________
Gnash-dev mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/gnash-dev
