- From: Ray Denenberg, Library of Congress <rden@loc.gov>
- Date: Wed, 28 Apr 2004 10:31:21 -0400
- To: <www-zig@w3.org>
From: "Mike Taylor" <mike@indexdata.com>
> I've been carefully reading what the on-line Z39.50 standard
> says about Scan, and there seem to be a couple of mistakes or
> ambiguities. I have no idea whether these issues are reflected in
> Z39.50-2003, as the HTML on the site is still from the 1995 version.
Z39.50-2003 is available via
http://lcweb.loc.gov/z3950/agency/document.html
or directly, http://lcweb.loc.gov/z3950/agency/Z39-50-2003.pdf
>
> In section 3.2.91 (Scan Service), towards the bottom of the page
> http://lcweb.loc.gov/z3950/agency/markup/05.html
> Table 12 shows the parameters of the Scan request and response. The
> "Number-of-entries" parameter is listed as being in the response,
but
> not the request; but the below indicates that it is in both:
Typo in the '95 version, the 'x' was omitted in the left box.
Corrected in 2003, see page 56.
> The ScanResponse ASN.1 at
> http://lcweb.loc.gov/z3950/agency/asn1.html#ScanResponse
> includes a parameter
>
> attributeSet [8] IMPLICIT AttributeSetId
>
> which does not seem to be documented in the prose. What is this?
Note that "list of suggested attributes" is documented (both in '95
and '03) the editor decided that the attribute set id didn't need to
be explicitly named in the service description.
> What the heck is suggestedAttributes all about? The prose says:
> A list of suggested attributes for use in subsequent Scan
> requests (useful for scanning multiple indices, e.g. author
> and title, at the same time).
You sent a scan request with a set of attributes, and the server is
telling you that you might get a better result with this set instead.
(Not trying to defend it, just trying to explain what it's about.)
> Why is alternativeTerm a SEQUENCE OF AttributesPlusTerm rather than
> just a single AttributesPlusTerm? Both the name of the ASN.1
element
> and the prose ("A suggested alternative term.") are clear that a
> single term is expected here.
It's supposed to be "one or more...."
> Has anyone, ever, in the history of the universe, implemented
> OccurrenceByAttributes? :-)
Les Wibberley!
Received on Wednesday, 28 April 2004 10:42:54 UTC