At 08:27 AM 22/02/2005 +0100, you wrote:

>Hi,
>
>Coming from a company (now defunct) that had a very specialized DB
>engine for extremely fast matching of thousands of criteria, I'd be
>interested to learn hos Firebird works in this regard and discuss if a
>similar technology could be implemented in firebird. This technology can
>match thousands of criteria on million record DB:s in fractions of a
>second, regardless of selectivity.
>
>What is the appropriate forum for this?

Not this one, please.

Actually, you've recently missed threads on this topic in both
Firebird-architect and Firebird-devel.

You could review these threads via the newsgroup interface and decide what
more you might like to add to the discussion. Actually, Firebird-Architect
is the right place for discussing what the engine might do; and
Firebird-tools to ask about what real-life people have
implemented. Keywords are "fuzzy" +"text" + "search"...

On Fb-devel, you might like to ask Nickolay to explain his changes to the
algorithms for the SQL string-matching predicates in Fb 2. That way, I
might be able to listen in and write a half-way decent explication of it in
the release notes. :-))

As far as the indexing technology is concerned, search the archives of this
and the other lists for Ann Harrison's postings. It comes up
frequently. You can also delve at www.ibphoenix.com for writings on the
topic.

For changes to indexing technology in Fb 2 (there are quite a lot!) get the
Fb 2 release notes, in draft, available for preview at
http://firebird.sourceforge.net/download/prerelease/

^heLen