Bug in affine2d documentation

here In the example, it is stated a ccw rotation is desired, but the actual tranform does a cw rotation because of the non-standard way the tranform is performed (multiply the matrix by a row vector instead of multiplying the column vector by the matrix):

 Define 10-Degree Rotation in the Counter-Clockwise Direction
 Create an affine2d object that defines the transformation.
 theta = 10;
 tform = affine2d([cosd(theta) -sind(theta) 0; sind(theta) cosd(theta) 0; 0 0 1]) 
 Apply forward geometric transformation to an input (U,V) point (5,10).
 [X,Y] = transformPointsForward(tform,5,10)
 X = 6.6605
 Y = 8.9798

They probably wanted to use the a transpose of the used (standard literature ccw rotation) matrix.

[Mariano]

Well, what I found is that, although according to the literature you are right (ie. this matrix should produce a CW rotation because it is trasposed), the Matlab function imwarp actually rotates the image in a CCW direction with this tform! This, I think, is even worst. It imeans that Matlab needs a wrong matrix (at least, different form the literature, for instance Gonzalez and Wood, or even wikipedia) to perform a correct transformation. I would appreciate a clarification from the Matlab team. Thanks!