L2/04-102

From: Dr. U.B. Pavanaja
Date: 2004-02-10 10:37:26 -0800
Subject: [indic] Bug in Kannada collation

There is a bug in the collation chart for Kannada
(http://www.unicode.org/Public/UCA/latest/allkeys.txt).
I am pasting the extracted Kannada table below -

0CBC  ; [.0000.0113.0002.0CBC] # KANNADA SIGN NUKTA
0C82  ; [.0000.0114.0002.0C82] # KANNADA SIGN ANUSVARA
0C83  ; [.0000.0115.0002.0C83] # KANNADA SIGN VISARGA
0CE6  ; [.0E29.0020.0002.0CE6][.0000.016F.0002.0CE6] # KANNADA DIGIT ZERO
0CE7  ; [.0E2A.0020.0002.0CE7][.0000.016F.0002.0CE7] # KANNADA DIGIT ONE
...snip...
0CCB  ; [.1750.0020.0002.0CCB] # KANNADA VOWEL SIGN OO
0CC6 0CC2 0CD5 ; [.1750.0020.0002.0CCB] # KANNADA VOWEL SIGN OO
0CCC  ; [.1751.0020.0002.0CCC] # KANNADA VOWEL SIGN AU
0CCD  ; [.1752.0020.0002.0CCD] # KANNADA SIGN VIRAMA
0CD5  ; [.1753.0020.0002.0CD5] # KANNADA LENGTH MARK
0CD6  ; [.1754.0020.0002.0CD6] # KANNADA AI LENGTH MARK

Kannada sign Avagraha (0CBD) has been placed wrongly between HA and LLA.
Since Avagraha is a sign, it should be placed in the beginning, along with
Nukta. I am not sure whether it should be placed before or after Nukta.
Both Nukta and Avagraha do not originally belong to Kannada. Nukta belongs
to Hindi and Avagraha to Sanskrit. I am not able to find any book where
both Nukta and Avagraha are used in some sort of collation. I am still
searching for one.

Regards,
Pavanaja
--------------------------------------------------------------------------
Dr. U.B. Pavanaja
CEO, Vishva Kannada Softech
Think Globally, Act locally