U+0F39 TIBETAN MARK TSA -PHRU blocks ligatures

[dscorbett]

Font

NotoSerifTibetan-Regular.ttf

Where the font came from, and when

Site: https://github.com/googlefonts/noto-fonts/blob/96e3d51ab4ad78e74a4cb0752dc25d010d6946de/unhinted/NotoSerifTibetan/NotoSerifTibetan-Regular.ttf Date: 2020-01-02

Font version

Version 2.001

Issue

U+0F39 TIBETAN MARK TSA -PHRU blocks ligatures.

Character data

དུ༹ར༹ྦཋ༹ྲ U+0F51 TIBETAN LETTER DA U+0F39 TIBETAN MARK TSA -PHRU U+0F74 TIBETAN VOWEL SIGN U U+0F62 TIBETAN LETTER RA U+0F39 TIBETAN MARK TSA -PHRU U+0FA6 TIBETAN SUBJOINED LETTER BA U+0F4B TIBETAN LETTER TTHA U+0F39 TIBETAN MARK TSA -PHRU U+0FB2 TIBETAN SUBJOINED LETTER RA

Screenshot

[simoncozens]

I'm seeing broken clusters in your example text, suggesting that TSA PHRU doesn't really want to go there. Should it go after the subjoined things instead? (དུ༹རྦ༹ཋྲ༹ has no dotted circles.)

[dscorbett]

Are you testing with HarfBuzz?

[simoncozens]

Ah, right. Harfbuzz is reordering, Safari/CoreText is not:

I don't know if that's relevant for the issue anyway.

[dscorbett]

In the first cluster, U+0F74 TIBETAN VOWEL SIGN U and U+0F39 TIBETAN MARK TSA -PHRU have different canonical combining classes, so their relative order should not matter.

In the last two clusters, the tsa ’phru follows the head consonant and precedes the subjoined consonant. The Unicode Standard, Version 15.0 says “Where used, the character U+0F39 TIBETAN MARK TSA -PHRU occurs immediately after the consonant it modifies.” I don’t know if it ever modifies these particular head consonants, but in principle, it is allowed to precede a subjoined consonant.

Therefore, Core Text is wrong.

[simoncozens]

I think this is now fixed; I'm getting the TSA -PHRU's on the ligatures: