RE: character sets: Term as OCTET STRING

Ray, I understand your point very well. This is typically a difference
between theory and practice. Of course you are right. But look it from my
point of view:

We have a Z39.50 client application. When this application leaves our
offices, we have no idea what is going to happen with it. We have to make
sure that it can talk to as many servers as possible from all kinds of
flavors. 

When I send Term as InternationalString, even if only after successful
Character Set Negotiation, I have a limited chance of success.
When I send Term as OCTET STRING, it will always be successful: I have never
seen a server that does not accept OCTET STRING.

OCLC is a rather important organization in our business. When a Geac client
cannot talk to an OCLC server because the Geac client sends
InternationalString and OCLC does not accept that (I think that Ralph said
that today), then the customer is not happy - and will probably put the
blame on us.


Pieter

> -----Message d'origine-----
> De : Ray Denenberg [mailto:rden@loc.gov]
> Envoyé : mercredi 10 juillet 2002 17:21
> À : zig
> Objet : Re: character sets: Term as OCTET STRING
> 
> 
> 
> Pieter Van Lierop wrote:
> 
> >  It is relatively easy to support InternationalString.
> >
> > But the question is: How long will it take before all 
> version 3 servers in
> > the world will accept InternationalString as Term type? And 
> the answer is,
> > that we will never be sure that all servers would want to 
> do that. So in
> > that case, for the sake of interoperability, when I am a 
> Z39.50 client and I
> > am sending a request to server I don't know, I use OCTET 
> STRING because that
> > gives the best chance on success.
> 
> But Pieter, consider this argument:  We're assuming a version 
> 3 server with
> character set negotiation. Thus some implementor took a 
> version 3 implementation
> (which did not previously support InternationalString) and 
> implemented character
> set negotiation on top. It's safe to assume the implementor  
> read the character
> set negotiation definintion. If he read it correctly he would 
> note that he has
> to implement InternationalString in order to properly 
> implement character set
> negotiation. And as you note above " It is relatively easy to support
> InternationalString."   If he has to modify the server 
> anyway, why didn't he
> implement it? Or to put it more bluntly, why would the so 
> implementor blatantly
> ignore the specification?
> 
> --Ray
>

Received on Wednesday, 10 July 2002 12:00:07 UTC