Overlapping below-base marks and vattu ra in Devanagari

[dscorbett]

Font

NotoSerifDevanagari-Regular.ttf

Where the font came from, and when

Site: https://github.com/googlefonts/noto-fonts/blob/46eb8fa0c46d16a87eee3e763baa3efb7423a8b4/phaseIII_only/unhinted/ttf/NotoSerifDevanagari/NotoSerifDevanagari-Regular.ttf Date: 2019-10-20

Font version

Version 2.000

Issue

On glyphs representing clusters with the circumflex-shaped below-base ra, the below-base anchor point is too high.

Character data

ङ्रॖछ्र᳘ U+0919 DEVANAGARI LETTER NGA U+094D DEVANAGARI SIGN VIRAMA U+0930 DEVANAGARI LETTER RA U+0956 DEVANAGARI VOWEL SIGN UE U+091B DEVANAGARI LETTER CHA U+094D DEVANAGARI SIGN VIRAMA U+0930 DEVANAGARI LETTER RA U+1CD8 VEDIC TONE CANDRA BELOW

Screenshot

[JelleBosmaMT]

The font files are generated from source files in Glyphs format. I know that in Summer 2019 the tools used by Google needed to be updated to correctly handle the anchors that composites are expected to inherit from base glyphs. The font file from the link is from 26 September 2019. I can reproduce the bug with the downloaded font file, but not with a font generated by Glyphs from the same source file.

It is possible that the software has improved in the mean time and the issue goes away when re-generating the fonts.

[JelleBosmaMT]

@marekjez86 We tried running the old source file through the current pipeline and it does not look good. For the GPOS mark to base lookup that deals with these low marks I compared the decompiled data of Glyphs.app and pipeline. Nine composite marks are missing. There are 124 fewer anchors, for composite glyphs. And where the number of anchors matches, they might be from the wrong component (I haven't tried to count them). It is unlikely an issue for Devanagari Serif only.

[nizarsq]

Tested the current font in this directory and anchor point lgtm in this font.

[simoncozens]

Yes, this is now fixed.