Writeup of last week's Geodata IRC Meeting
jo at frot.org
Tue Aug 15 23:09:38 PDT 2006
Feedback on the above would be very much welcomed on the wiki.
This meeting revolved around discussion of the Simple Catalog
Interface idea that's been on the lists for a while. Stefan Keller
(who couldn't make the meeting) has been summarising his thoughts
particularly in regards to an OAI-PMH compliant interface for geodata
Chris jumps in to show off a quick prototype of an OpenLayers
interface on top of a store of WMS services and what's advertised in
their GetCapabilities. http://world.freemap.in/metadata/maplayer.html
- this is with the as-yet-unreleased OpenLayers 2.0 with box drawing
support among many other things, with GetCapabilities collected from
the Google Search API.
Underneath it is a small webservice powered by Sean Gillies' OWSLib -
also in heavy redevelopment - which emits JSON objects which
correspond to some part of the capabilities - extents and so on.
Underneath that is the database that Jo's been working on to store the
Geodata Metadata Requirements worked out over previous geodata
meetings for OSGeo's Geodata Repository project.
Jo's been working with some of Stefan's ideas - a very simple OAI-PMH
interface for transferring geometadata around in RDF. It sort of
works, but needs more metadata underneath to really show its face.
Norman is focused on development of a lightweight discovery mechanism
- "a bit like a service registry, but we do all the finding";
something that can feed into a distributed caching effort. The goal
state is a server that allows JPIP streaming from multiple data (
service ) resources. He's building a J2EE service that re-uses the
OSGeo geometa schema.
Jo talks about the background of the OSGeo effort - how a metadata
store connects to the early-days Geodata Discovery Working Group and
how this can lead into dynamic repackaging of data. Schuyler notes
that these are basically two different problems with the same
underlying solution - a simplest-useful-thing metadata store with one
or more commonly understood RESTful interfaces.
"nhv is intrigued by the possibility of the process of getting data
avail for a service might just be the process whereby the metadata is
formalized" - this should work just as well for data one is publishing
as data one is discovering. It's in the same kind of space as mapdex,
but with uses for data providers as much as data aggregators.
There basically is no decision or agreement on what the RESTful
interface should look like - in a way it doesn't matter as long as we
can agree on it - or what it transfers as long as people can easily
reuse it - and this is the easiest part of the problem - the core is
getting the model right, and hopefully Geodata Metadata Requirements
is a long way towards that. Jo and Norman will continue to work on their
implementations and socialise them on the geodata mailing list. We'll
hopefully have a BOF or breakout session on the subject at FOSS4G.
More information about the Discuss