| Commit message (Collapse) | Author | Age |
| |
|
|
|
|
| |
The copyright message now found in the source files is correct.
|
|
|
|
|
| |
Due to changes in
https://github.com/jamesgk/ufo2ft/commit/6c46f7050bfdd346c33c1312f07830c48fcc07f2
|
|
|
|
|
| |
This is hopefully the last time we have to do this. In response to
https://github.com/googlei18n/cu2qu/commit/56f36a1b2ab673c25ac81756d5c251909a7d41e2
|
|
|
|
|
| |
cu2qu is a new library without a currently well-defined API, so these
calls have changed and will probably change again.
|
|
|
|
|
| |
The UFOs store contours in ccw order, and the TrueType spec seems to
say they should be cw. The old binaries are cw.
|
|
|
|
|
|
| |
This is a small change which passes the threshold at which dots will
have eight points instead of four, which makes a noticeable difference
in shape.
|
| |
|
| |
|
| |
|
|
|
|
| |
It's not much, but maybe a start....
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is (hopefully) a temporary hack until a more robust solution to
the issue is found.
Basically the problem is that RoboFab pens do float-to-float
comparisons to decide whether to remove duplicate points from
contours, and breaks with BooleanOperations output (which can include
point coordinates with very small fractional components). The
italicizing code assumes certain behavior from RoboFab pens and can't
handle duplicate points being kept.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
If components of components aren't also decomposed, we may miss some
contours in the resulting glyphs.
|
|
|
|
|
|
| |
Apparently this method returns a different font than we expect in
decomposeGlyph, probably due to the glyph copying we do when mixing
masters. The easiest thing to do is just not trust or use the method.
|
|
|
|
|
|
|
|
|
| |
Without this change, some of the master data contains int values
which are not interpolated correctly due to numpy array behavior.
For example adding float values to a numpy array of ints will
always result in an array of ints, so a scale value which should be
a float between 0 and 1 might instead be rounded.
|
|\ |
|
| | |
|
| |
| |
| |
| |
| |
| |
| | |
I'm not sure why we weren't doing this before, or why there is even a
check to prevent adding top or bottom anchors to glyphs without
unicode values. But this change seems appropriate, particularly since
it fixes #69
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This is necessary because of some implementation details in RoboFab;
RAnchor objects always evaluate to False in a boolean context.
Fixes a bug in which "top" anchors were not being added to composite
glyphs, since they are added only when the "parenttop" anchor is
present (which was checked for with a statement like "if anchor:").
|
|\ \
| | |
| | | |
Correct metadata related to font names and weights.
|
| | |
| | |
| | |
| | | |
Without this, ufo2fdk sets both Light and Regular to 400 by default.
|
| | |
| | |
| | |
| | |
| | | |
This propagates to name ID 6 in the name table, and font name info in the
CFF table.
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This uncomments some code which distinguishes between Regular/Bold and other
weights when setting the UFOs' styleMapFamilyName attribute, which is used by
the FDK to set the output OTFs' family names (this association is documented
here: http://unifiedfontobject.org/versions/ufo2/fontinfo.html). It's not clear
why this code was commented out in the first place.
Part of #37
|
| | |
| | |
| | |
| | |
| | | |
Basically a redo of 9292364c1a122498ada2616fdd20683c27c80e15, except more
conservative.
|
| | |
| | |
| | |
| | |
| | | |
This reverts commit 9292364c1a122498ada2616fdd20683c27c80e15. Necessary
because it caused condensed glyphs to italicize incorrectly.
|
| | |
| | |
| | |
| | |
| | |
| | | |
Updated glyphs require more kerning subtable breaks to avoid crashing
makeotf. The subtable sizes are still decided somewhat arbitrarily based on
trial-and-error.
|
| | | |
|
|\ \ \
| | |/
| |/| |
|
| | | |
|
| |/
|/|
| |
| |
| |
| | |
Doing this conversion visibly changes the appearance of some composite glyphs,
since e.g. a component scale of 1.6 would be rounded to 2. What doesn't make
sense is why this was working with FontLab....
|
| |
| |
| |
| | |
This caused some glyph mixes to fail.
|
| |
| |
| |
| | |
This affects the creation of condensed masters, without it they had no kerning info.
|
| |
| |
| |
| | |
This was causing problems with Thin Italic font generation.
|
|\ \
| | |
| | | |
Fix bug causing inaccurate anchor coordinates.
|
| | |
| | |
| | |
| | | |
This has now been fixed upstream.
|
| | |
| | |
| | |
| | | |
Ideally we could call RGlyph.move(offset, anchors=False), but we have to bring back "shiftGlyphMembers" from the original toolchain because RGlyph.move does not work as documented.
|
| | | |
|
| | |
| | |
| | |
| | | |
These issues are solved with the updated anchor resources.
|
| | |
| | |
| | |
| | | |
This is necessary for correctly adding anchors in the immediately following code.
|
| | |
| | |
| | |
| | | |
I introduced this bug in aa285dc861fb8a5b5083e1de3bb00ea1ef658cb1 while adapting updated scripts from Paratype.
|
|/ / |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
| |
| |
| | |
With anchors fixed, we shouldn't need this check. Rollback of 55519e636b119dcdd91319364c572d3550050a43.
|