GSUB lookup needs a fix to handle akhand sequences in Telugu, Kannada

[arjuna rao chavala <arjunaraoc googlemail com>]

I have been attempting to fix the bug  for akhand sequences in Telugu (https://bugzilla.gnome.org/show_bug.cgi?id=579398  Lohit Telugu, Pothana fonts) and noticed the problem in Kannada (https://bugzilla.gnome.org/show_bug.cgi?id=604060 Problem for both Lohit kannada and Kedage) Surprisingly the second bug does not appear for Lohit Telugu.  Lohit Telugu and Pothana (telugu) work fine on windows platform.

The akhand sequences seem to work fine with Hindi (Lohit Hindi font).

Akhand only needs to be applied for Ka+Virama+Sha and not when ka+matra(dependent vowel)+sha  (Telugu test cases.:  see cited bug for more details)

Akhand should not be applied for ka+virama+ka+sha (Test case for Kannada , see cited bug for more details).

There is an interplay between shaper and the font, and it has become very difficult to debug this. Really need help from people more familiar with GSUB code, on how the lookahead  string length is determined in Pango and  how the character level  interpretation information  is passed to the GSUB code.

BTW, KDE and OpenOffice do not have the problem with Telugu bug that was
 mentioned.

As this bug is fundamental affecting the proper rendering of multiple languages, I request immediate help.

Regards
Arjun
http://tech4society.blogspot.com
http://teluginux.blogspot.com (Telugu)