Subject: Re: Revisit: Merge GILS Attribute Set into Bib-1
Eliot Christian (echristi@usgs.gov)
Date: Thu, 12 Feb 1998 17:22:20 -0500
Message-Id: <3.0.3.32.19980212172220.007723c0@130.118.4.2> Date: Thu, 12 Feb 1998 17:22:20 -0500 To: gils@cni.org From: Eliot Christian <echristi@usgs.gov> Subject: Re: Revisit: Merge GILS Attribute Set into Bib-1 In-Reply-To: <01BD37CF.458A4750@fay.turner@nlc-bnc.ca>
On 2/12/98, Fay Turner <fay.turner@nlc-bnc.ca> wrote:
>
> If the proposal is accepted, and I hope it is, will there be a coordinated
> approach for systems switching over to the use of the new Bib-1 numbering?
I agree that there will be a transition hurdle to get over. I think the
goal during the transition is that clients can send either or both use
attribute numbers with a query, and servers accept either or both.
Here's my sense of the impacts of renumbering for the various situations
involving clients and servers that currently support search using the
Bib-1 use attribute numbers, including all GILS-compliant servers and
GILS-aware clients.
If your client or server has not been distinguishing the attribute set,
then you have already been treating GILS attributes as though they were
in Bib-1. Just add the new Bib-1 use attribute numbers to your
available use attribute number table. Clients can send either or both
with a query, and servers should be prepared to accept either or both.
If your client or server has been distinguishing attribute sets, but not
supporting the GILS attribute set, again you can just add the new Bib-1
use attribute numbers to your available use attribute number table.
If your server has been distinguishing attribute sets, and supporting
the GILS attribute set, you can also just add the new Bib-1 use
attribute numbers to your available use attribute number table. As
clients make the transition to the new Bib-1 use attribute numbers, you
will see fewer and fewer search requests with the GILS attribute set
specified.
If your client distinguishes attribute sets and supports the GILS
attribute set, you have a more complicated situation. The key is to
know whether the target server supports Z39.50 version 2 or version 3,
and whether the target server/database supports the GILS attribute set.
Since we do not yet have a deployed mechanism to negotiate profiles, you
have to either ask the Explain database (not widely deployed), ask the
searcher (probably doesn't know and surely doesn't expect to be asked),
or just "guess". I'll not presume to describe the "guess" approach but
it seems to me you would send both possible use attribute numbers as
terms with an "or" between them.
In addition to client and server implementors (in which I include
gateways), there are also builders of metadata collection mechanisms.
For these folks, I expect the renumbering is transparent unless the
database ingest procedure has been designed to expect use attributes
numbers as unique tags. I would think the same applies to folks who
build usage guidelines and cross-walks among metadata formats.
Of course, we also have to address the set of folks who build metadata
schemes.
For instance, the GEO profile inherits attributes from GILS and CIP
inherits attributes from GEO. With the GILS attributes in Bib-1, this
inheritance chain would merely shorten by one link. During transition,
the attribute numbers tables should be updated to associate to either
attribute set/number. For the STAS and CIMI Profiles, I would think
having the GILS Attribute numbers in Bib-1 would simplify
interoperability.
Please do share with the list any comments on the transition, as well as
any other unforeseen complications surrounding this proposal.
Eliot Christian, US Geological Survey, 802 National Center, Reston VA 20192
echristi@usgs.gov Office 703-648-7245 FAX 703-648-7112 Home 703-476-6134
This archive was generated by hypermail 2a16 : Tue Mar 23 1999 - 03:55:41 EST