Re: Octet Strings and utf-8

Mike Taylor wrote:

> PLEASE not more random semantics shoehorned into access points!  If we
> really want to go this route (though I'll be interested to see what --
> if any -- responses Ralph's question turns up), ...

I'll assume for the sake of moving this along that there won't be any
responses to Ralph's question; still, I'm not prepared to for Z39.50 to
declare itself a text-only protocol.

However, after giving this more thought I think there is a reasonable
rationalization path, if we can tolerate a certain level of convolution.

The key is, anyone who wants Z39.50 to support non-text searching (e.g.
boiling point using IntUnit)  should be told that they need to support
version 3 (I don't think this is an unreasonable demand, since for example
you can't use IntUnit as search term without version 3). We can then amend
the character set negotiation definition to say that if version 2 is in
force then negotiation applies to octet string when used as the search
term.

--Ray

Received on Friday, 15 March 2002 09:54:30 UTC