(Fwd) Re: [Indic-computing-standards] Re: Malayalam Half-U: ho

[Dr. U.B. P. <pav...@vi...>]

------- Forwarded message follows -------
From:           	"Dr. U.B. Pavanaja" 
<pav...@vi...>
To:             	Keyur Shroff 
<key...@ya...>
Date sent:      	Thu, 14 Nov 2002 09:51:22 +0530
Subject:        	Re: [Indic-computing-standards] Re: 
Malayalam Half-U: how
Priority:       	normal

> 
> --- "Dr. U.B. Pavanaja" <pav...@vi...> wrote:
> > From these discussions I can infer one thing: We need a 
> > mechanism of choosing one of the many possible display forms for a
> > particular combination. 
> > 
> > We are having a similar requirement for Kannada for the case of
> > "arkavattu" (reph) and "half ra". Both forms of display are
> > possible and both are correct. I had mentioned this to the people
> > responsible for OpenType specifications of Indic scripts.
> > Currently they don't have any plans to do this changes.
> 
> Can't it be done using Zero Width Joiner (U+200D)? ZWJ is used to
> produce half forms of consonant/ligature in Indic scripts.
> 
> (Ra Halant ZWJ) sequence will produce desired half form of Ra.

I should have expalined what I meant. I had 
explained this in 
the Sep.15-16 workshop at Blore. Let me explain 
again.

In Kannada, the combination of "ra + halant + <any 
consonant 
other than ra>" can be displayed in two forms -
1) With the second consonant in full form followed 
by the 
arkavattu (reph) character. This method has come 
from Samskrit 
and not original to Kannada. Nowadays majority 
people use this 
form.
2) With the full form of "ra" followed by the half 
form (vattu 
form, below base form) of the second consonant. This 
is original to
Kannada. All combinations of consonant + halant + 
consonat use this
method.

In our Nudi keyboard there is a provision to select 
any one of 
the form at our wish.

In case of OTF, the display form is decided by the 
logic put 
into the font (GPOS, GSUB, etc). We can not decide 
on the 
display form based on the keyboard input.

I was asking for such a provsision in OTF 
specifications. 
Probably, if such a provision is made in OTF specs, 
then the 
problem faced by Malayalam in case of chillus might 
be solved. I am
not very sure of this. May be some Malayalee can 
throw light.

> > 
> > Another point I would like mention here: The sorting rule in
> > Unicode has got nothing to do with the character code pages. They
> > are different. Unicode has two charts -character chart and the
> > collation table. Details of collation are available at
> > www.unicode.org/tr10
> 
> Yes, separate algorithm is required. Otherwise it will lead to wrong
> sorting sequence. I'll try to explain this by giving simple example.

I fully agree. The collation table has a section for 
every 
language and not for the script. This helps for 
Hindi, Samskrit,
Marathi etc who have the same script, but have 
different sorting
rules.

We have done similar in case of Kannada. We use 
ISCII for 
storing the text. There is a provision in Nudi to 
save as ISCII. For
sorting we follow the pure consonant approach by 
knocking off the
implicit "a" in every consonant. That means we 
consider the consonants
as "k, kh, g, gh" rathar than "ka, kha, ga, gha". We 
have arrived at
an encoding table for sorting purpose called KSCLP 
(Kannada Script
Code for Language Processing). The problem of half-
consonants (as in
rajkumar and rajakumar) is solved by this approach.

Rgds,
Pavanaja--------------------------------------------
---------
Dr. U.B. Pavanaja
Editor, Vishva Kannada
World's first Internet magazine in Kannada
http://www.vishvakannada.com/

Note: I don't worry about pselling mixtakes

------- End of forwarded message -------------------
-----------------------------------------
Dr. U.B. Pavanaja
Editor, Vishva Kannada
World's first Internet magazine in Kannada
http://www.vishvakannada.com/

Note: I don't worry about pselling mixtakes