[PLUG] Devanagari unicode: composite letters ( जोडाक्षरे ); latex

Mayuresh mayuresh at acm.org
Fri Jan 15 08:27:27 PST 2010


For an example of the issue I am trying to understand, try typing the word 
शुक्रवार (Marathi for Friday in case you can't see devanagari properly) in 
devanagari.


You can see the applications like word processors, email editors, browsers 
etc. (most of them) render the word properly.

However other applications like terminal emulators (say konsole or gnome- 
terminal) split the second letter i.e. क्र as "half ka" and "ra".

It's obvious that there is no separate unicode assigned to (and it's not 
feasible too) to each such composite letter. So it is up to the application to 
compose these letters correctly.

Apart from Devanagari, even Arabic seems to have the same issue (as per 
wikipedia explanation of unicode).

Isn't it strange to leave such core things as font rendering to applications 
rather than them being at system level? Can't there be a common system-wide 
component to do that?


Mayuresh.

PS: This all originated from my attempt to use devanagari in a latex document. 
I was able to do it using xelatex and unicode package - but for issues with 
letters like above. Basically xelatex doesn't have the above mentioned logic 
to render some of those special composite letters. (Hence it's painful to see 
it being so application dependent.)

Will appreciate help in getting letter क्र (as an example) rightly rendered via 
latex. (I know other methods like generating pdf by export from OOffice etc. 
That's fine. The question is more about latex.)


More information about the plug-mail mailing list