[JDEV] Non english messages problem (solution)

Alexander Tsvyashchenko ndl at ndl.unicyb.kiev.ua
Mon Apr 9 15:01:32 CDT 2001

Hello, "wil at home" <wil at dready.org>!

Yes, I understand that setting assumed encoding for all ICQ transport rather then for every message CAN produce errors. The main idea is these errors should be quite unlikely events. Your example with cp1251/koi8-r about ICQ transport is somewhat wrong: ALL russian/ukrainian users of ICQ have to use cp1251, otherwise noone can understand them (Linux ports of ICQ simply convert messages from koi8-r to cp1251 before sending and vise versa), so setting assumed encoding cp1251 for ICQ transport is a good choise in this case. I think similar situation exists also for other languages/encodings. Yes, I understand there may be some problems, but it seems that this scheme is a good point to start. It is not so hard to implement this and it will be rather effective for many (maybe even most) of users. Later it may be useful also to add second variant (with selecting encoding on client), but for start it would be very nice if even first variant would be implemented.

	Good luck!                                     Alexander.

On Mon, 9 Apr 2001 06:21:32 +0800
"wil at home" <wil at dready.org> wrote:

WAH> You have a valid point that it places a lot of effort on the client side to
WAH> do charset conversion.  However, that is the most complete solution, simply
WAH> because it puts the power at the users hands.
WAH> If the conversion is done on the ICQ transport, e.g.
WAH> Hackee has 2 friends, A and B.  A uses linux with koi8-r encoding, while B
WAH> uses Windows with cp1251 encoding.  If the preferred encoding is set to
WAH> cp1251, and Hackee receives a message from A, the ICQ transport would
WAH> convert the raw bytes from ICQ, treating it like it was cp1251, and the end
WAH> result would be wrong.  How is Hackee gonna reverse engineer it, switch the
WAH> encoding for the ICQ agent and tell A to resend?
WAH> Another way would be to associate each user with a preferred encoding, that
WAH> sort of solves the problem, but still could be frustrating if the sender
WAH> decides to switch OS :)
WAH> wil

More information about the JDev mailing list