Re: [Gnash-dev] Re:Matrix math (strk)

[Mark Voorhies]

On Wednesday 03 September 2008 16:38:56 strk wrote:
> On Wed, Sep 03, 2008 at 09:40:27AM -0700, Mark Voorhies wrote:
> > On Wednesday 03 September 2008 04:10:54 strk wrote:
> > > On Tue, Sep 02, 2008 at 11:23:01PM -0700, Mark Voorhies wrote:
> > > 
> > > > I think that pushing the reflection into sx gives the expected 
behavior as 
> > > > long as the sx != 0 (choosing sx rather than sy is arbitrary).
> > > > 
> > > > For the reflection case, atan2 returns the wrong sign (the logic for 
> > assigning 
> > > > quadrants assumes only rotation), so it is still necessary to do the 
sign 
> > > > correction in get_rotation:
> > > > 
> > > > E.g.
> > > > 
> > > > double
> > > > matrix::get_x_scale() const
> > > > {
> > > >     if(determinant() < 0)
> > > >     {
> > > >        // Capture reflection in x-scale
> > > >        return -sqrt(((double)sx * sx + (double)shx * shx)) / 65536.0;
> > > >     }
> > > >     else
> > > >     {
> > > >        return sqrt(((double)sx * sx + (double)shx * shx)) / 65536.0;
> > > >     }
> > > > }
> > > > 
> > > > double
> > > > matrix::get_y_scale() const
> > > > {
> > > >     return sqrt(((double)sy * sy + (double)shy * shy)) / 65536.0;
> > > > }
> > > 
> > > Using sqrt here would make it impossible for Y scale to be negative.
> > > 
> > > --strk;
> > > 
> > 
> > Correct.  If x_scale and y_scale are both positive then there is no 
> > reflection.  If x_scale xor y_scale is negative then there is a reflection 
> > (about the y-axis or x-axis respectively).  If both x_scale and y_scale 
are 
> > negative then there is no net reflection; instead, the reflections about 
both 
> > axes yield a rotation, which is redundant to the rotation component.
> > 
> > Since the rotation and translation components _can't_ describe a 
reflection, 
> > we have to capture it as a negative value in one of the scale components.  
I 
> > arbitrarily chose x_scale, which should work in all cases except for 
x_scale 
> > = 0.  (x_scale = 0 xor y_scale = 0 will give a zero determinant and 
collapse 
> > the image to a line.  In this case, capturing a reflection as a 180 degree 
> > rotation _will_ work.  If both x_scale and y_scale = 0, the image will 
> > collapse to a point and both reflection and rotation become moot).
> 
> Do you think you can produce an SWF testcase showing a failure in Gnash ?
> See testsuite/misc-ming.all/matrix_test.c for sample code (can add to it).
> The .getBounds() calls there triggers matrix transformation of the shape
> bounds (a rectangle). None of those test fail. get_x_scale/get_y_scale
> are NOT involved in those operations, and we're debating whether they 
> should in general...
> 
> --strk;
> 

Trunk is building successfully on my computer now, so I should be able to play 
with the testsuite over the weekend.  I'll post to the list again when I have 
a better understanding of the problem and some informative testcases.

--Mark