Noto Serif Tibetan deletes U+25CC DOTTED CIRCLE in certain contexts

[dscorbett]

Font

NotoSerifTibetan-Regular.otf

Where the fonts came from, and when

Site: https://github.com/googlefonts/noto-fonts/blob/4d7396a4fae475975b367e225f2e8f49292d8837/unhinted/otf/NotoSerifTibetan/NotoSerifTibetan-Regular.otf
Date: 2022-04-17

Font version

Version 2.102

Issue

Noto Serif Tibetan deletes dotted circles in some contexts. If the input string contains a dotted circle, the font should not delete it. One context is specifically in ⟨ཧྲཱིཿ◌࿆⟩. Another context is specifically in ⟨རྒྱ྄◌ུ⟩. Another context is between certain multi-letter single-glyph stacks and certain below-base combining glyphs, as in ⟨ཀྐ◌ྑ⟩ or ⟨ཀྱ◌ཱུ⟩.

If this is necessary to work around the limitations of certain shaping engines, you could trying moving the rules from 'blws' to 'blwf'. 'blwf' is scoped to the cluster whereas 'blws' applies to the whole buffer. If the dotted circle is inserted by the shaping engine, the whole thing will be one cluster so 'blwf' will apply, and if the dotted circle is part of the input string, there will be multiple clusters so 'blwf' will not apply.

I can’t find any information about ⟨རྒྱ྄ུ⟩. Is that an attested cluster? Why is it just that specific cluster, but not any other clusters containing U+0F84 and a vowel sign?

Character data

ཧྲཱིཿ◌࿆
U+0F67 TIBETAN LETTER HA
U+0FB2 TIBETAN SUBJOINED LETTER RA
U+0F71 TIBETAN VOWEL SIGN AA
U+0F72 TIBETAN VOWEL SIGN I
U+0F7F TIBETAN SIGN RNAM BCAD
U+25CC DOTTED CIRCLE
U+0FC6 TIBETAN SYMBOL PADMA GDAN

རྒྱ྄◌ུ
U+0F62 TIBETAN LETTER RA
U+0F92 TIBETAN SUBJOINED LETTER GA
U+0FB1 TIBETAN SUBJOINED LETTER YA
U+0F84 TIBETAN MARK HALANTA
U+25CC DOTTED CIRCLE
U+0F74 TIBETAN VOWEL SIGN U

ཀྐ◌ྑ
U+0F40 TIBETAN LETTER KA
U+0F90 TIBETAN SUBJOINED LETTER KA
U+25CC DOTTED CIRCLE
U+0F91 TIBETAN SUBJOINED LETTER KHA

ཀྱ◌ཱུ
U+0F40 TIBETAN LETTER KA
U+0FB1 TIBETAN SUBJOINED LETTER YA
U+25CC DOTTED CIRCLE
U+0F71 TIBETAN VOWEL SIGN AA
U+0F74 TIBETAN VOWEL SIGN U

Screenshots