<p>Norman,<br>[...]<br>> Unfortunately I know very little about postgis (though I have wanted<br>> to use it for a while), so when the schema further develops I would be<br>> interested to see how you index getcapabilities (free text search?
<br>> spatial index?)</p>
<div>Instead of writing persistence layer from scratch you could take a look at this <a href="http://jpoxspatial.zones.ch/">http://jpoxspatial.zones.ch/</a> .</div>
<div> </div>
<div>In <a href="http://geometa.info">geometa.info</a> we investigate a combination of unstructured and structured data, so we use both, free text (with unbeatable Lucene) an a simple schema - surely too simple compared to work done here by Tom with OWSCat, Paul with uDig and others (start here:
<a href="http://www.digitalearth.com.au/2006/06/23/lightweight-web-resource-catalogue/">http://www.digitalearth.com.au/2006/06/23/lightweight-web-resource-catalogue/</a>).</div>
<div> </div>
<div>But my main interest here is not to discuss back end schematas or functionality of (search) service providers but the least required extension/specialization of Dublin Core information model to be exchanged (currently its 'coverage' and 'protocol') in a most established RESTful protocol for metadata harvesting (my currently evaluation ranks OAI-PMH, WFS profiled then CSW profiled). Here metadata includes data about data as well as data about services (here: WxS which contain data).
</div>