> Date: Mon, 22 Apr 2002 19:53:06 +1000
> From: Alan Kent <[log in to unmask]>
>
> After the wild fury of email on the Explain-- [...]
Huh? What fury?
> [...] I have come to the opinion that there are two possible
> directions for implementing CQL which relates to some of the
> opinions about what indexed field names to use for searching.
>
> One approach is to say 'I want to write queries without knowing the
> structure of the database'. In this case, attribute sets like Bib-1
> are important as they are standardised.
>
> The other approach is to say 'queries will only be written when the
> structure of the database is known.' That is, the database structure
> would be interrogated (using Explain or whatever) when parsing a CQL
> query.
> [...]
> Do people
> (1) Agree with my claim to there being two directions?
Yes.
> (2) Think that one is more important than the other?
Yes. In every anti-Z39.50 opinion I have ever heard expressed, the
one thing that _always_ comes up as its redeeming grace is the culture
of abstraction in Z39.50, and particularly the way in which this
facilitates interoperability and multicasting. I think it would be a
tragedy for SRW to throw that out.
Obvious example: if CQL includes (for example) DC.title, then a
multicasting client can spit the same title-search out to a whole
batch of servers rather than having to hand-craft the search for each
one.
> left truncate, stem, greater than, first in field, GEO regions
> overlap, etc relate to attribute lists. Do field names in queries
> identify complete attribute lists (as recommended in the Bath
> profile)? If so, what 'modifiers' are allowed? Do we hard code in
> ">" (greater than) for example, or do we keep it very
> generic.
The way I remember Ralph's original (admirably concise) CQL document,
these issues were already addressed: truncation (and therefore
stemming) with charcter masking; relations with the use of <, >=,
etc.
Thing is, I can't find that paper anywhere now: it doesn't seem to be
linked from the MA page, google doesn't know about it, and Ralph's
page at OCLC is also mysteriously quiet on the subject :-)
So does this mean that the document has been withdrawn?
_/|_ _______________________________________________________________
/o ) \/ Mike Taylor <[log in to unmask]> www.miketaylor.org.uk
)_v__/\ "We apologise for the delay. This is due to various delays
due to the train in front of us being delayed" -- genuine,
real British Rail announcement.
|