[cairo] correct arc_to error bound calculation

[Bertram Felgenhauer]

Hi,

here is a patch that does three things:
1st. it splits out the major axis calculation from cairo-pen.c
2nd. it corrects a mistake in that calculation.
3rd. it makes arc_to_error base its calculation on this major axis
calculation instead of the bogus use of Eigen values.

comments:
1st. This seems to be better than just copying the code.
2nd. The two nondiagonal entries of the transformation
matrix had to be swapped to match the derivation above
3rd. The Eigen values are not even real numbers in most cases,
namely as soon as a rotation is involved in the transformation.

The Eigen value calculation has no users after this patch is
applied and could be removed.

Bertram Felgenhauer
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cairo-arc-to-use-major-axis.diff
Type: application/octet-stream
Size: 488 bytes
Desc: not available
Url : http://lists.freedesktop.org/archives/cairo/attachments/20050814/15c1aac0/cairo-arc-to-use-major-axis.obj