Discussion ensued on the problems of efficiently retrieving small snippets of XML from a large database of snippets.

Relational database implementations seem to slow down when the query you are doing against your datastore involves complicated (XPath) evaluations on a column, we guess because each row has to be retrieved/parsed/evaluated (a more expensive version of the like operator).

Some RDBMS vendors can provide speed up on this evaulation by creating temporary tables to represent your xml document, but this seemed to provide a maintainence overhead.

So if anyone reading this knows of a product that can search a large datastore of small XML documents with speed and ease of maintainability, let us know in a comment, ta.

Some other topics covered were:

• JavaGroups (free multicast comms)
• TimesTen (commercial realtime product)
• problems with distributed caches
• some background on Apache Geronimo
• creative commons for software
• ...and 'single-photon detectors and their application on quantum cryptography' (you can tell a couple of physicists turned up :-) )

Oh and star spotting turned up Tom Quinn (Matthew MacFadyen) from BBC's 'Spooks' but he didn't join us for a chat, obviously working on a very hush, hush java project for the government...

Another fun night out, from the java.meetup site, cheers all.