Remove PUA codepoints from glyphs.

[guidoferreyra]

Many glyphs are using PUA codepoints, we should remove it and double check are accesible through features.

[djrrb]

I agree we should put them in features and make sure the correct codepoints are used when possible, but I do hesitate to break backwards compatibility unless you think it’s required. @guidoferreyra can you confirm?

To have a record of the glyphs currently using PUA:

Fractions

Only two of these have proper encodings. The frac feature handles other arbitrary fractions. Should we leave these using PUA?

• F506 oneseventh: also encode as u+2150 ⅐
• F507 twosevenths
• F508 three sevenths
• F509 foursevenths
• F50A fivesevenths
• F50B sixsevenths
• F50C oneninth: also encode as u+2151 ⅑
• F50D twoninths
• F50E fourninths
• F50F fiveninths
• F510 sevenninths
• F511 eightninths

Ligatures

• F001 fi: move to U+FB01 fi
• F002 fl: move to U+FB02 fl

Proportional figures

Implemented via OpenType feature pnum, no encoding necessary. (Will also rename glyph name suffix to '.lf' for clarity.)

• F800 zero.alt
• F801 one.alt
• F802 two.alt
• F803 three.alt
• F804 four.alt
• F805 five.alt
• F806 six.alt
• F807 seven.alt
• F808 eight.alt
• F809 nine.alt

Superiors and Inferiors

Use unicode subscripts and superscripts instead or in addition.

• F80A zero.sups
• F80B one.sups
• F80C two.sups
• F80D three.sups
• F80E four.sups
• F80F five.sups
• F810 six.sups
• F811 seven.sups
• F812 eight.sups
• F813 nine.sups
• F814 zero.sinf
• F815 one.sinf
• F816 two.sinf
• F817 three.sinf
• F818 four.sinf
• F819 five.sinf
• F81A six.sinf
• F81B seven.sinf
• F81C eight.sinf
• F81D nine.sinf

Logos

Omitting from font.

• F200 ubuntu (wordmark)
• E0FF uniE0FF (logo)
• F0FF uniF0FF (superscript logo)
• F000 uniF000 (hinting glyph)

[guidoferreyra]

The font has other "breaking" changes to already consider it as new major release, so my opinion is the following:

Fractions: I don’t see what is the point of having those precomposed glyphs for very rare fractions. But if we need to keep them I think it’s better to remove the PUA to not alter user’s character string when activating the feature.

Ligatures: I agree

Proportional figures: I agree. +1 for the renaming too

Superiors and Inferiors. To me just using the superior and inferior unicode values will be enough, no need to add PUA

[djrrb]

Ok sounds like a good plan, I will implement this shortly.

[djrrb]

Changed in e16eed91024b8f645f306c2ac6e17ee5307ea784

It’s a bit unusual because the encoded zerosuperior was set to the numr feature and the unencoded zero.sups is set to the sups feature. But I don‘t see a problem with that.