actually Adobe has described the issue in the FP9 release notes:
http://www.adobe.com/support/documentation/en/f…

Transform Matrix transformations are not reflected in respective MovieClip/DisplayObject properties. Properties like scaleX, scaleY, and rotation are not changed as the result of changes to a DisplayObject’s transformation matrix (flash.geom.Transform, flash.geom.Matrix).

they suggest to use Matrix-Transformations instead to solve the scaleX / scaleY function.

which i found solved in the “fl.motion.MatrixTransformer” – which again you can find in the Flash CS3 IDE path:
[Program Files]AdobeAdobe Flash CS3deConfigurationActionScript 3.0Classesflmotion

which i found solved in the “fl.motion.MatrixTransformer” – which again you can find in the Flash CS3 IDE path:
[Program Files]AdobeAdobe Flash CS3deConfigurationActionScript 3.0Classesflmotion

trace(a.scaleX, a.scaleY, a.rotation, a.transform.matrix);
trace(b.scaleX, b.scaleY, b.rotation, b.transform.matrix);
trace(c.scaleX, c.scaleY, c.rotation, c.transform.matrix);

0.5 0.5 180 (a=-0.5, b=0, c=0, d=-0.5, tx=125, ty=173)
-0.5 -0.5 0 (a=-0.5, b=0, c=0, d=-0.5, tx=250, ty=173)
-0.5 -0.5 180 (a=0.5, b=0, c=0, d=0.5, tx=375, ty=173.45)

Scaling x or y by -1 is equivalent to flipping on that axis. Scaling x and y by -1 at the same time is equivalent to rotating by 180 degrees. Decomposing a matrix into scale and skew/rotation requires an interpretation that is ambiguous, at times.

For a and b, the matrices are identical in scale. However, the interpretation appears to be swayed by the whether the scaleX and scaleY properties are set. Flash seems to cache scaleX and scaleY, rather than doing a fresh decompose every time. That's my guess.

c is interesting because it is both doubly negative and rotated, which cancel out.

trace(a.scaleX, a.scaleY, a.rotation, a.transform.matrix);
trace(b.scaleX, b.scaleY, b.rotation, b.transform.matrix);
trace(c.scaleX, c.scaleY, c.rotation, c.transform.matrix);

0.5 0.5 180 (a=-0.5, b=0, c=0, d=-0.5, tx=125, ty=173)
-0.5 -0.5 0 (a=-0.5, b=0, c=0, d=-0.5, tx=250, ty=173)
-0.5 -0.5 180 (a=0.5, b=0, c=0, d=0.5, tx=375, ty=173.45)

Scaling x or y by -1 is equivalent to flipping on that axis. Scaling x and y by -1 at the same time is equivalent to rotating by 180 degrees. Decomposing a matrix into scale and skew/rotation requires an interpretation that is ambiguous, at times.

For a and b, the matrices are identical in scale. However, the interpretation appears to be swayed by the whether the scaleX and scaleY properties are set. Flash seems to cache scaleX and scaleY, rather than doing a fresh decompose every time. That’s my guess.

c is interesting because it is both doubly negative and rotated, which cancel out.