Glenn, I studied the OpenType spec. With that knowledge I assumed, perhaps naively, that for all OT fonts a correct rendering can be obtained by following the directives in the tables in the font.
The code of the ComplexScripts branch provides specific processing of each Indic font (until now Devanagari). The main action seems to be that each word is first syllabized and then each syllable is separately submitted to the substitution code. Do I understand correctly that the font tables alone do not allow one to achieve a correct rendering, and that it is necessary to understand the syllabic structure of the text? Simon