Some SignWriting diacritic sequences do not overlay U+1D9FF SIGNWRITING HEAD

[dscorbett]

Font

NotoSansSignWriting-Regular.otf

Where the font came from, and when

Site: https://github.com/googlefonts/noto-fonts/blob/2b1fbc36600ccd8becb9f894922f644bff2cbc9b/unhinted/otf/NotoSansSignWriting/NotoSansSignWriting-Regular.otf
Date: 2020-12-31

Font version

Version 2.001

Issue

Some SignWriting characters in Unicode are diacritics that apply to U+1D9FF SIGNWRITING HEAD. When no base is available, U+25CC DOTTED CIRCLE is inserted. That mostly works, but there are some diacritic sequences that get spacing glyphs. They should not get spacing glyphs.

One example is <U+1DA84, U+1DA9F>. When applied to U+1D9FF, the diacritic appears next to the base instead of overlaid on it.

Another example is <U+1DA5E, U+1DAA7>. The U+1DA5E is applied to U+1D9FF correctly, but the U+1DAA7 is rendered with a dotted square as if it were invalid. The intended rendering of this diacritic sequence only appears when it does not follow a base U+1D9FF.

This is a general problem that affects more than those two examples.

Character data

𝧿𝪄𝪟
U+1D9FF SIGNWRITING HEAD
U+1DA84 SIGNWRITING LOCATION HEAD NECK
U+1DA9F SIGNWRITING FILL MODIFIER-6
𝧿𝩞𝪧
U+1D9FF SIGNWRITING HEAD
U+1DA5E SIGNWRITING TONGUE MOVES AGAINST CHEEK
U+1DAA7 SIGNWRITING ROTATION MODIFIER-8
𝩞𝪧
U+1DA5E SIGNWRITING TONGUE MOVES AGAINST CHEEK
U+1DAA7 SIGNWRITING ROTATION MODIFIER-8

Screenshot

[dscorbett]

This bug is still present in version 2.005, though it has changed a bit. Here is how my three original example strings are rendered in version 2.005:

The first string is a head with a combining mark. It should look like one grapheme cluster, but it looks like two.

The second string is a head with a combining mark modified by a rotation modifier. It should look like one grapheme cluster, but it looks like two.

The third string is a combining mark modified by a rotation modifier. There is no base. Inserting a dotted circle is appropriate. I think it would be even better if the combining mark appeared on the dotted circle, instead of on a separate head, because that makes it look like the combining mark is applied to a base. Compare <U+1DA00> (another string of a mark without a base) where the mark appears on a dotted circle instead of on a head next to a dotted circle: