[ckan-dev] Conventions for CKAN data in the webstore

[Rufus Pollock]

As Friedrich reported last week the webstore is operational (see
<http://wiki.ckan.net/Webstore> for more info).

We now need to work out some conventions for how to store data from
datasets registered on a CKAN instance into a webstore.

Here's a first proposal. It is motivated by the use case of
automatically putting data into the webstore so that we an API
available as per CREP 4 [1]. I have also put it on the wiki here:
<http://wiki.ckan.net/Webstore#Convention_for_Storing_CKAN_Datasets>

* There is a designated user for the CKAN instance. Suggest using the
CKAN site_id variable (which is usually the domain name). So for
thedatahub.org the username would be thedatahub.org. (Note: given
current authz for webstore.datahub.org based on ckan this means you
need a thedatahub.org user on thedatahub.org ...)

* Each resource being converted gets its own database. Name the
database after the resource id.

* Store this info back onto the resource in your CKAN instance.
Suggest following 'spec' for an attribute field on resource metadata
named 'webstore':

 {
    url: {url for that db}
 }

E.g. for a resource

 url: {resource-url}
 format: ...
 ....
 webstore: {
    url: "http://webstore.thedatahub.org/thedatahub.org/0000-aaaa-bbbb-dddd"
 }
 ...

Regards,

Rufus

[1] http://trac.ckan.org/ticket/1190