< snip >
> I believe that you have explained his observation. He has not
> committed the numerically entered rotation and it is negated when he
> attempts to rotate the image with the mouse, and is much more
> noticable at larger angles of rotation.
I would like to add that I never actually do the process mentioned here
when I want to correct the image by rotation I just wanted to see if I
could reproduce what had been reported.
Gimp-user mailing list