quick run through the status of the repository project espec. re metadata

[Jo Walsh]

dear all,

This is an email I wrote the other day when I started one on
#telascience then got too lagged out to contribute to the discussion.

- It would be of benefit for the repository project to have a mentor, and 
  I have recently neither made time for it or been able to find anyone with 
  the time to take it on. I'm making more time now.

(This goodwill-overload seems endemic to OSGeo right now - committed
volunteers are max capacity - and we are all here because we're
*already* volunteering on open source and open data projects...
and it can be hard to find paths in to helping that don't just involve
taking up the time of the people who are already helping.) But I have
renewed energy for the repository and want to be more active in
helping to set up and coordinate.

- we *did* get as far as setting up a simple web-based metadata
  annotator at FOSS4G, on the .216 box which answers to public.geodata.osgeo.org 

It should be enough to enable us to use to document what is being
collected at telascience both for SDSU and for OSGeo. There is no
authentication in it but I will just LIMIT POST to basic auth and not
worry about it. This is to say that in the short term this is a viable
fix to get us all onto the same page, to sync up the metadata model
with the data in the world and work from there. I definitely want to
work with GeoNetwork as well espec. as the project is on the verge of
entering OSGeo incubation for proper - hooray!

- I had so many people speak to me at FOSS4G with interest in
  contributing data or of hosting a node of a distributed effort. 

I think that what the collective-we here can most usefully provide is
best practise, a kind of blueprint for setting up a data archive with
open source and open standards components. 

- I'm sorry to hear that there still looks like so much uncertainty
  over what the OSGeo metadata model is and what we are requireing
  from contributors. 

As i see it http://wiki.osgeo.org/index.php/Geodata_Metadata_Requirements 
is a good discussion of the baseline though it covers a lot more than that
now. 

http://wiki.osgeo.org/index.php/Geodata_Metadata_Model looked to me complete 
- perhaps it did not cover licensing issues, everything else basic. If
there are serious pieces missing from the above we should know about it!

As far as 'required' vs 'optional' properties goes - it is my own
personal judgement call and i *believe* this reflects the collective
irc judgement call, that almost nothing should be 'required' to be
able to create an entry - very basic things like a data set or a
contact person has to have a name. Beyond this the data is just on a
sliding scale of possible metadata completeness - it could be a
percentage of the fields we have, and until a certain threshold is
reached we won't republish the metadata records via OAI-PMH,
wfsbasic/owscat, what have you. I hope that this makes sense - that we
have a set of "strenously encouraged" fields - as little should be
required as is possible.

Right now this model is implemented as a PostGIS schema with a
python object interface to that and a web based editing client and
even a commandline interface though it is just a sketch. It should be
real easy in theory to XSLT transform the data out towards FAO
GeoNetwork, so that we can "leverage" the rigourous implementation of
CSW-ebRIM and other unmentionables, and do our own lighter weight
interfaces (owswfscatbasic, oai-pmh) as well. 

I personally am more excited about seeing the same application maybe
work as a meta-level indexer, something that is missing now, part of a
distributed catalog service and also getting into feature level
metadata and more useful / readable ways to DescribeFeatureType. But I
won't keep letting this distract me from the basic application. 

I think that the geodata repository project would benefit most from a
few people being able to take a dedicated span of time to be able to
do processing, documentation, work on the toolkit. I got as far as
sketching out a fantasy funding application (with Google in mind, as
it happens, modelled on an application written by a friend who works
there, for development work on OpenSSL...)

http://wiki.osgeo.org/index.php/Fantasy_Geodata_Funding_Application -
given the context it has a software bias which is maybe getting
carried away. I would be happy to receive any expansion or critique
and interested to hear from anyone who might want to get involved in
this. The format requires a goodish explanatory paragraph of what would 
be done and achieved by it for each person - and more of an intro about 
what the broader aims are and why this would be worth supporting. I'm up 
for putting in another burst on this and trying to shop it around other
orgs - on the understanding that some tithe would go to the
Foundation. I ran this by a few people six weeks or so ago, got no
serious eek! noises, and have not thought of it since. So this list
incidentally describes what work i think is to be done before the
geodata repository effort could be called officially bootstrapped...

Well, this has been a rambling braindump written in response to
specific questions on irc. I hope it has supplied more clarity than
obscurity. 

cheers,


jo