I have doubts about the Indic_Positional_Category (InPC) values proposed for four new dependent vowels being added in Unicode 10.0.0.
On examining the vowel chart (p1265 of http://www.unicode.org/Public/10.0.0/charts/CodeCharts.pdf) one may feel quite comfortable with assigning the property values: 1143E..1143F ; Top # Mn [2] NEWA VOWEL SIGN E..NEWA VOWEL SIGN AI 11440..11441 ; Right # Mc [2] NEWA VOWEL SIGN O..NEWA VOWEL SIGN AU However, on consulting Section 3.6 of Anshuman Pandey's 'Proposal to Encode the Newar Script in ISO/IEC 10646' http://www.unicode.org/L2/L2012/12003r-newar.pdf, one finds that after the seven headless consonants GA, NYA, TTHA, NNA, THA, DHA and SHA, the dependent vowels take forms more appropriate to the property values 1143E ; Left # Mn NEWA VOWEL SIGN E 1143F ; Top_and_Left # Mn NEWA VOWEL SIGN AI 11440..11441 ; Left_and_Right # Mc [2] NEWA VOWEL SIGN O..NEWA VOWEL SIGN AU Now, I have no idea what the effect of a right-to-left directional override should be on the combining marks, but in general I believe gc=Mn makes more sense for U+1143E and U+1143F, so I am not challenging that property assignment. However, I do wonder what the best property values are for a renderer, such as Microsoft's Universal Shaping Engine. It seems to me that it may be better to start with the properties involving 'Left' and use contextual substitutions to convert the dependent vowels to components of the correct position. However, this does seem more complicated than the general decomposition of multipart vowels. In particular, for headed consonants, a glyph substitution is required to replace the head by a part of the vowel symbol; the default glyph will not be appropriate. It is entirely possible that a font will simply replace a headed consonant and any of these four vowels by a ligature glyph, leaving reordering to be considered only for the seven headless consonants. Has this matter been considered? If so, is the rationale recorded anywhere? Richard.
