- From: Henrik Dahl <hdahl@inet.uni2.dk>
- Date: Wed, 17 Nov 2004 16:42:26 +0100
- To: "'Archie Warnock'" <warnock@awcubed.com>
- Cc: <www-zig@w3.org>
Hi Archie (I hope to be more compliant now). Thanks for you positive opinion. Yes, it has the idea of working smoothly in parallel when available and falling back on more plain behaviour with more plain servers. Best regards, Henrik Dahl -----Oprindelig meddelelse----- Fra: Archie Warnock [mailto:warnock@awcubed.com] Sendt: 17. november 2004 16:13 Til: Henrik Dahl Emne: Re: SV: InitResponse Options Bit 9 Henrik Dahl wrote: > Mr. Warnock, That's my father. I'm just Archie ;-) > Have you taken 3.2.2.1.2 point 1 in Z39.50-1995 into consideration? > > In many cases it may be highly contextually dependant which databases may be > searched at the same time. I think this is the correct approach: > 1) You should just try to search the various databases by the same search > request. > 2) If the server doesn't support that, it should send a non-surrogate > response record for the diagnostic with code 23 back to the client. This > diagnostic is defined on page 86 and has this meaning: "Specified > combination of databases not supported". > 3) If the client receives such a diagnostic it must search each database > individually. > > In this way you just use common features for your specific request. > > What do you think of this kind of approach? I'll pass this suggestion on to the group developing the specification. Actually, I like the approach quite a bit in that it allows the clients to assume that searching multiple databases in one search request is available by default. It does generate a bit of frustration if one has to guess at what combinations are available, but I think we can discuss that issue in terms of specific implementation. -- Archie -- Archie Warnock Internet: warnock@awcubed.com -- A/WWW Enterprises http://www.awcubed.com -- As a matter of fact, I _do_ speak for my employer.
Received on Wednesday, 17 November 2004 15:44:22 UTC