This reduces the amount of skyline input coming from glyph outlines by
2x.
benchmark for arguments: input/regression/mozart-hrn-3
raw data: {'ff2061b518': [2.72, 2.69, 2.71], 'e8212e15e2': [2.7, 2.68, 2.69]}
Version ff2061b518: Create skylines directly from glyph outlines.
e8212e15e2 - Split glyph contours in up/down segments for skylines
med diff -0.020000
med diff -0.738007 % (e8212e15e2 is faster)
benchmark for arguments: -I carver MSDM
raw data: {'ff2061b518': [51.83, 51.99, 52.05], 'e8212e15e2': [51.27, 51.2, 51.48]}
ff2061b518: Create skylines directly from glyph outlines.
e8212e15e2 - Split glyph contours in up/down segments for skylines
med diff -0.720000
med diff -1.384882 % (e8212e15e2 is faster)
rebase
https://codereview.appspot.com/569700043
contours
https://codereview.appspot.com/569700043
Diff:
Patch does not apply to current master.
rebase
https://codereview.appspot.com/569700043
Patch does not apply to current master.
rebase
https://codereview.appspot.com/569700043
rebase; drop experimental work
https://codereview.appspot.com/569700043
Fails make check
The log file referenced just gives the same errors as the output above.
jonas.
https://codereview.appspot.com/569700043
Passes make, make check and a full make doc.
Reg test diffs here (patch 7) ~ 6MB!
https://cloud-u8zj2dc4b.yourownnet.eu/s/qJNCHJzFsmrRM3R
looks like the bugfixes caused minor formatting diffs. I'll separate out the bugfixes so we can be sure.
rebase on bugfix change
https://codereview.appspot.com/569700043
Patch does not apply to current master