[Indic-computing-devel] UTF-8

[Tapan S. P. <ta...@ya...>]

First, let me apologize for my previous post for two reasons 1) the 
ibook im on seems to have encoded nl's funky in my attachment, and 2) i 
forgot to delete the rest of the mail when i replied.  at least i can 
promise (2) wont happen again...

Next, I wanted to comment on this exchange...

<quote>
-- You can also have FontCode->UNICODE tranlation or
vice-versa. but that better be at the backend and
not at the client side.

-- I agree. Because UNICODE would double the traffic between the server 
and the client.
</quote>

Now i agree there is a strong argument that this should be done server 
side b/c of portability, but it is not always true that Unicode 
transfers should take 2x as long.  This is only currently true b/c the 
current standard Unicode one-byte encoding (UTF-8) is biased towards 
latin scripts, in that latin chars take up one byte and all others 2-3.  
One could very easily imagine and implement an encoding that would be 
biased towards indian scripts, in that iscii chars would take up only 
one byte and all others 2 or 3.  Now Im not saying we should do this, 
and obv there is the issue of how to distinguish between diff scripts, 
and related issues in unicode round trips, but its something to think 
about...

Regards,
Tapan


_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com