Canonically equivalent Myanmar strings look different

[dscorbett]

Font

NotoSansMyanmar-Regular.ttf
NotoSerifMyanmar-Regular.ttf

Where the font came from, and when

Site: https://github.com/googlefonts/noto-fonts/blob/34e98229863e627d0f841e124a8657d5d0348b04/phaseIII_only/unhinted/ttf/NotoSansMyanmar/NotoSansMyanmar-Regular.ttf
Site: https://github.com/googlefonts/noto-fonts/blob/34e98229863e627d0f841e124a8657d5d0348b04/phaseIII_only/unhinted/ttf/NotoSerifMyanmar/NotoSerifMyanmar-Regular.ttf
Date: 2019-06-09

Font version

Version 2.001

Issue

After bases with descenders, the sequences <U+1037, U+103A> and <U+103A, U+1037> are rendered differently, although they are canonically equivalent. I think the latter has the intended rendering.

Character data

ဈ့်ဈ့်
U+1008 MYANMAR LETTER JHA
U+1037 MYANMAR SIGN DOT BELOW
U+103A MYANMAR SIGN ASAT
U+1008 MYANMAR LETTER JHA
U+103A MYANMAR SIGN ASAT
U+1037 MYANMAR SIGN DOT BELOW

Screenshot