On Jan 31, 2007, at 4:53 PM, Steve Christensen wrote:
On Jan 31, 2007, at 9:11 AM, David Duncan wrote:
On Jan 31, 2007, at 07:58 AM, Mike Kluev wrote:
Wait, what's wrong with CGAffineTransformInvert?
Hadn't thought of that, but same mathematical problems without
all the code maintenance ones.
Doing this is akin to trying to draw bicycle spokes by appending
a rotation to the context. By the time your done, the floating
point errors have accumulated to the point that your spokes don't
look correct.
Tho, isn't an inverse just a straight mathematical operation on
the actual values in the transform matrix? If so, there's
shouldn't be an accumulation error, but I suppose there could be
rounding ones where you end up with a 1.000000000000001 or
0.9999999999985 someplace.
Define "straight mathematical operation." If you mean each entry in
the inverse matrix is found by operating on one element of the
original, then no, it isn't. Google for "inverting a matrix."
