W3C home > Mailing lists > Public > www-zig@w3.org > December 2002

Re: Version 1.0 of SRW and CQL

From: Eliot Christian <echristi@usgs.gov>
Date: Mon, 02 Dec 2002 05:00:13 -0500
Message-Id: <5.1.0.14.0.20021202043327.012d59e0@gsvaresm03.er.usgs.gov>
To: www-zig@w3.org, Doug Nebert <ddnebert@usgs.gov>

At 11:24 AM 11/29/2002 +0000, Robert Sanderson wrote:

>[...]
> > profile communities are using Sebastian's PQN syntax as the query
> > format for Z39.50 URL's and for Web Services, including UDDI. PQN,
>
>This would have been interesting to know during the development of SRW and
>CQL.

I believe I did raise this as an objection to what was then a "CCL push".

> > I see that ZOOM supports PQN/PQF too.
>
>Only in as much as it has a polymophic query object.  It also supports
>CQL, Cheshire's implementation of CCL and so forth. Just because YAZ is
>commonly used doesn't make PQN the only way to express type 1 queries :)

Granted. My point is that PQN is the closest thing we have to a "friendly"
representation of Z39.50 query structures that also preserves an essential
feature of Z39.50--its abstraction of Use and Relation attributes. The
use of numeric identifiers for well-known attributes is also especially
handy in gaining wide consensus on syntax. (It is virtually impossible
to get broad consensus on names for attributes, and we may soon have
trouble getting consensus even on use of Latin-1 to convey such names.)

> > It appears to me that CQL does not support the large set of numeric
> > Use Attributes needed for Geo (or STAS), nor the set of Relation
> > Attributes needed for Geo. Wishing to avoid fragmenting of the Z39.50
> > community, is it possible to reconcile the ZING work with the need
> > for support of PQN in the query syntax?
>
>Yes :)
>Use attributes and their attribute architecture cousins are easily defined
>as new index sets in CQL.  These can be done per community without
>registering anything, unlike the OIDs required for Z39.50.

Sure, but the whole point of interoperability is to expose commonalities
and that requires a registry. Whether a attribute "non-inteligible
identifier" is an OID or some other token should not matter really.

>[...]
>Relations are more difficult as they require changing the CQL spec.  We
>did discuss GEO relations during the CQL discussions but decided that as
>none of us were experts we should leave it to the geospatial community to
>tell us what their needs are in this regard

I suspect that few people in the Geo community bother to track the ZIG
list.

>[...] Mike Taylor I believe already has a CQL library in Java that
>converts the CQL format into the equivalent PQN for building gateways.

That's good but to enable use of SRW/SRU I think we need agreements and
tools for making CQL out of arbitrary PQN, don't we? (Noting that PQN
itself directly translates from/to Z39.50 query structures.)

Eliot



Eliot Christian echristi@usgs.gov  1-703-648-7245 FAX 1-703-648-7112
US Geological Survey, 802 National Center, Reston VA 20192
Received on Monday, 2 December 2002 05:06:57 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 29 October 2009 06:12:23 GMT