Ufo2ft

- Raise error when multiple glyphs claim the same Unicode codepoint. When expanding color layers to glyphs, ensure any codepoint is stripped to prevent unintentional clashes (739).
- Use ``fontTools.cu2qu`` instead of standalone ``cu2qu`` module, no changes, just one less dependency (756).
- Added compile option for static/variable TTF outputs to ``dropImpliedOnCurves`` to omit on-curve points that are halfway between two consecutive off-curve points, slightly reducing glyf table size with no visible differences (753).
- [constants] Updated Indic / USE script lists (755).
- Require fonttools >= 4.40.0

- [featureWriter.ast] Have iterClassDefinitions search not just top-level but inside feature and lookup blocks as well (732).
- Added ``autoUseMyMetrics=True`` compile option (for TrueType output only) to allow disabling the automatic setting of ``USE_MY_METRICS`` glyf component flags, which is not needed for fonts without hinted metrics (741).
- [kernFeatureWriter] Only use exported glyphs when generating mark filtering sets (740).

- Dropped support for Python 3.7 which reached EOL. We now require Python >= 3.8.
- Take into account extra substitutions found in DesignSpace rules when classifying glyphs by various unicode properties in feature writers (731).
- Added support for compiling TrueType instructions and flags stored in the UFO's ``public.truetype.*`` lib keys (654).
- Automatically compute COLRv1 ClipBoxes if none are explicitly provided. This is enabled by default for static builds only. You can set colrAutoClipBoxes=True for interpolatable/variable builds if safe to do so (728).
- Improved "no default source" error (677).
- Warn about mis-cased `` Automatic Code`` insertion marker comments; also warn if there are no insertion markers but kerning.plist is non empty and there are manual kern/dist features in the FEA code (719).
- Respect openTypeNameRecords overrides for the predefined name IDs (727).
- Reorder class definitions in feature writers such that they appear before the lookups that use them (725).
- Renamed dottedCircleFilter module to dottedCirlce, was wrongly named; old kept for backward compat (716).
- [kernFeatureWriter] Avoid overwriting classDefs (705).
Register kern lookups for all defined languages (711).
Correctly split mark and base kerning pairs (713).
Don't skip over spacing marks when kerning (720).
- Fixed FEA include dir detection when in current dir (689)
- Automatically decompose components when 2x2 transforms differ (698)
- [kernFeatureWriter] Split kerning by script, not horizontal direction, which should be faster to compile and produce smaller GPOS. The old kernFeatureWriter is still kept in a separate module (679, 636, 697).

- Dropped support for Python 3.7 which reached EOL. We now require Python >= 3.8.
- Added support for compiling TrueType instructions and flags stored in the UFO's ``public.truetype.*`` lib keys (654).
- Automatically compute COLRv1 ClipBoxes if none are explicitly provided. This is enabled by default for static builds only. You can set colrAutoClipBoxes=True for interpolatable/variable builds if safe to do so (728).
- Improved "no default source" error (677).
- Warn about mis-cased `` Automatic Code`` insertion marker comments; also warn if there are no insertion markers but kerning.plist is non empty and there are manual kern/dist features in the FEA code (719).
- Respect openTypeNameRecords overrides for the predefined name IDs (727).
- Reorder class definitions in feature writers such that they appear before the lookups that use them (725).
- Renamed dottedCircleFilter module to dottedCirlce, was wrongly named; old kept for backward compat (716).
- [kernFeatureWriter] Avoid overwriting classDefs (705).
Register kern lookups for all defined languages (711).
Correctly split mark and base kerning pairs (713).
Don't skip over spacing marks when kerning (720).
- Fixed FEA include dir detection when in current dir (689)
- Automatically decompose components when 2x2 transforms differ (698)
- Copy old KernFeatureWriter from ufo2ft 2.30.0 into separate module (697)

Make it possible to kern Hiragana and Katakana against each other (683)

* Split kerning by script, not direction (679). This is a rework of 636 and should now properly split up kerning into per-script lookups without losing any. The major upside is that multi-script fonts with a lot of kerning will compile faster and smaller, because lookups will overflow less often and become denser.
* As a side-effect, kerning will now also be generated such that the shaper will actually apply it when the designer did not explicitly declare an appropriate `languagesystem` in the font's features. Before, kerning for Greek glyphs was effectively dropped unless the designer declared `languagesystem grek dflt;`.
* Another side-effect is that kerning between some common glyphs like `increment` and `asterisk` will now always be applied, where before due to circumstances it might have been stuck in e.g. the `LTR` instead of the `Common` lookup and therefore not applied if the script wasn't set correctly.