uvsDict uses list of tuples instead of dict

[moyogo]

FontTools cmap uses list of tuples instead of a dict for the UVS mapping.

[anthrotype]

public.unicodeVariationSequences is a dict of dict.. but I suppose that doesn't matter

lgtm

[anthrotype]

wait, why have the tests in #495 not failed?

[moyogo]

wait, why have the tests in #495 not failed?

Not sure. But I'm hitting another issue when compiling:
https://github.com/fonttools/fonttools/blob/main/Lib/fontTools/ttLib/tables/O_S_2f_2.py#L210-L216

	def updateFirstAndLastCharIndex(self, ttFont):
		if 'cmap' not in ttFont:
			return
		codes = set()
		for table in getattr(ttFont['cmap'], 'tables', []):
			if table.isUnicode():
				codes.update(table.cmap.keys())

The cmap subtable format 14 doesn't have a cmap attribute but isUnicode() returns True so this raises and AttributeError.

[anthrotype]

Not sure.

probably we're only testing values after built, but before actually compiling the table (in TTFont.save())

The cmap subtable format 14 doesn't have a cmap attribute but isUnicode() returns True so this raises and AttributeError.

oh, then it might need special handling. Could you PR that in fonttools?

[moyogo]

OK. The UVS test has been updated to crash if cmap.compile() fails. TIL the spec wants them sorted.

The keys are also hex strings instead of ints, so now that works properly.

For the cmap attribute, fontTools just adds a self.cmap = {} for format 14, so ufo2ft does the same.

[anthrotype]

thanks