Re: Sort criteria in Search Request

I'm joining this discussion late, as I  was away all last week......

Sebastian Hammer wrote:

> ..... I feel that concatenation is too heavyweight
> and clumsy a mechanism for the job.
>
> I think my preferred mechanism would be an extension to the search PDU,
> negotiated by an option bit. Runner-up would be an AdditionalSearchInfo, ....

I'm disturbed by the suggestion to use anything other than encapsulation. It
calls into question the entire ZIG/Maintenance Agency relationship and process.

Significant effort went into the encapsulation specification. Nevertheless, if
it's too heavyweight and clumsy then let's get rid of it, rather than continue
to ignore it.

All of the clarifications, amendments, etc, are being rolled into a consolidated
draft-for-ballot (that will be ready for review before the ZIG meeting, and will
be on the agenda for discussion at the meeting) of a proposed new
maintenance-revision of Z39.50.   The Encapsulation amendment will be part of
this revision. Those of you who do not see encapsulation as appropriate for
sort-lookahead should be either be prepared to explain why encapsulation should
remain part of the revision if it cannot be used with sort, or should be
prepared to argue that Encapsulation should be removed.

Now, assuming that encapsulation is potentially useful, it's important to
remember that it was aimed directly at the kind of functionality were
discussing. More specifically, the aim of encapsulation is optimization and
there are three types listed in the spec (see
http://lcweb.loc.gov/z3950/agency/amend/encapsulation.html) one of which is
"lookahead" characterized by a "Sort encapsulated within a Search, allowing the
server to optimize the search by knowing, a priori, the desired sort order".

Here is a suggestion: we could come up with an implementor agreement that
described how to use encapsulation explicitly for this purpose. It's just a
matter of sticking a sort pdu into the otherInfo parameter of a search. The
overhead of using encapsulation is negligible. We could even define an option
bit for this (thus you could negotiate the use of encapsulation exclusively for
sort-lookahead).  And we could possibly approve this at the October meeting and
get it into the revision.

Is this a good idea?


--
Ray Denenberg
Library of Congress
rden@loc.gov
202-707-5795

Received on Tuesday, 4 September 2001 11:55:42 UTC