On Wed, Jun 29, 2005 at 11:40:08AM +0200, Alex S wrote:
> Dot is defined to be x*x + y*y, but in this file the author defined
> cross to be x*x + y*y.
> And also cross should be x*y - y*x, but the author defined Dot to be x*y
> - y*x.
Thankfully, neither that Dot nor Cross macro is used in current code,
so I've simply removed both macros.
Incidentally, libnr/nr-point-fns.h has a dot function with the correct
definition, and a cross function that arguably gives the wrong sign,
depending on what you expect, e.g. whether you think y increases upwards
or downwards, and whether you think of cross in terms of dot product
with rotation in a certain direction or in terms of the basic scalar
operations. (I'm assuming that the conventional definition of (a cross b)
for two-dimensional vectors is ax * by - ay * bx.)
By Inkscape's angle conventions & its rationale (see below) (and with the
above assumption as to the traditional mathematical definition), the cross
function should be changed:
# Angle direction in Inkscape code: If you use the traditional
# mathematics convention that y increases upwards, then positive angles
# are anticlockwise as per the mathematics convention. If you take the
# common non-mathematical convention that y increases downwards, then
# positive angles are clockwise, as is common outside of mathematics.
This convention is intended to reduce problems with interfacing with
libraries & humans that use the mathematical convention.
Now that fred (the author of that cross function and most or all of its
callers) is no longer working on Inkscape, I might get around to doing a
mass change some time after 0.42 is released; though it's not very high
priority of course.