I would highly recommend not accepting this pull request just yet, as I would like to discuss the changes.
Entry translation is rotated first before performing the translation, with edir weighting occurring first, as we translate a rotated frame.
edir is a bit of a mess at the moment, as I found that for the rotation multiplications, edir needs to be taken into consideration beforehand due to the independence of edir on ds. This leads to too many and ugly uses of edir.
translate weights everything, so we only need to provide lw.
PTC uses path length for the calculation of rtbar.
Translation needs to occur on exit independent of algn.trn as rtbar can create a translation.
mang needs to be used for all rbends and sbends (most likely anything with an angle)
Comparison w/ ptc is difficult as the order of rotations used is x, y, s, not y, -x, s (rotating y and x simultaneously causes relatively high errors.)
I would highly recommend not accepting this pull request just yet, as I would like to discuss the changes.
edir
weighting occurring first, as we translate a rotated frame.edir
is a bit of a mess at the moment, as I found that for the rotation multiplications,edir
needs to be taken into consideration beforehand due to the independence ofedir
onds
. This leads to too many and ugly uses ofedir
.translate
weights everything, so we only need to providelw
.rtbar
.algn.trn
asrtbar
can create a translation.mang
needs to be used for all rbends and sbends (most likely anything with an angle)x, y, s
, noty, -x, s
(rotatingy
andx
simultaneously causes relatively high errors.)