Re: Z39.50 character encoding

At 14:07 21-02-2002 -0500, Ray Denenberg wrote:

>(a) Assign an option bit for utf-8 encoding.
>(b) Define an attribute for the encoding of a search term.
>(c) Do both.

I think I would be worried about doing both because of the added 
complexity. Of the two proposals, I favor (a) because I think it seems 
simpler, and my sense is that it will allow the cleanest fall-back for 
existing applications... Client suggests UTF-8, the server either 
rejects/ignores it, or approves. It's simple and clean, and as you suggest, 
it covers both search & retrieval. If you add a new attribute, you will be 
up against a large installed base of servers that react wildly 
unpredictably when faced with unknown attributes (or types). Many servers 
don't even check the attribute set OID, much less look at any attribute 
types they don't care about (the Bath profile & friends are changing this 
poor behavior, but it's a slow process). The Init option negotiation 
mechanism is beautiful in its simplicity -- it tends to work as you expect 
even when talking to a really dumb server.

So put my vote down for (a).

--Sebastian
--
Sebastian Hammer, Index Data <http://www.indexdata.dk/>
Ph: +45 3341 0100, Fax: +45 3341 0101

Received on Thursday, 21 February 2002 16:47:34 UTC