[NTG-context] Malayalam shaping with mkiv/LuaTeX

Rajeesh K Nambiar rajeeshknambiar at gmail.com
Thu Jan 9 22:34:49 CET 2014

On Thu, Jan 9, 2014 at 7:11 PM, Hans Hagen <pragma at wxs.nl> wrote:

>>> Remove \usemodule[font-odv], the code is part of the core and reloading
>>> it causes only problems.
>> Indeed, that fixes the compilation! And 2-part matras are fine now.
>> But the output (shaping) is not what is expected as you have observed.
>> Notably:
>> 1.  When a pre-base matra follows the "akhn" ligatures are broken
>> 2. "pref", "pstf" and "blwf" ligatures are not formed
> you need to identify the place in the code where that has to happen after
> which i can have a look at it (with real minimal examples)
With the initial patch I had sent, these were working okay (ie, akhn,
pref, pstf and blwf ligatures were being formed).
Minimal character sequence for test case:
\char"0D26 \char"0D4D \char"0D26 \char"0D47    %% <DA, VIRAMA, DA, E>
needs to form <ligature DDA,E>

With initial patch, in "methods.dev2", (syllablestart,syllableend)
were: (0D26, 0D26) {cluster DA, VIRAMA, DA} and (0D47,0D47) {E}.
Current beta gets (0D26,0D4D) {cluster DA, VIRAMA} and (0D26,0D47)
{DA, E}. Something changed in "analyze_next_chars_two", may be?

Okay - found that you had not considered 0D4D (virama/halant) not as
"halant{}" but as "stress_tone_mark{}". Adding 0D4D to the array
"halant{}" restores the previous behaviour - ligatures are formed but
with issues mentioned in the initial mail. "analyze_next_chars_two"
indeed looks for halant.



More information about the ntg-context mailing list