[okfn-discuss] urls as contracts in CKAN

Mon Nov 23 20:07:42 UTC 2009

2009/11/20 Rufus Pollock <rufus.pollock at okfn.org>:
> 2009/11/20 James Casbon <casbon at gmail.com>:
>> Hi,
>>
>> I was chatting to someone in the http://www.intermine.org/ team.
>>
>> They generate what are essentially views on their database which are
>> assigned unique urls (called templates in intermine terminology).
>> They want to know if another service is using this url, so that it can
>> become a 'contract' between two services.
>>
>> Should I encourage them to register these 'contract' URLs in CKAN?
>
> I don't see why not though it would be good to have a bit clearer idea
> of what exactly these urls will represent, how persistent they are and
> how many of them get generated.

I am using contract to denote a url that returns a certain result set
(say, all genes in drosphilia).  Publishing this contract would be a
commitment to preserve this result set through underlying model
changes and upgrades.

>
> It seems to me that the desire to 'cache' a given query (and perhaps
> even to considerit as a dataset in itself) is an interesting one and
> likely to become ever more frequent as we have larger single stores of
> data. Just a few weeks ago I had a conversation with Tony Hirst where
> he was suggesting something similar to this for complex SPARQL queries

Yes, this could be a secondary concern.  The primary requirement is to
publish and discover these endpoints, since they want to use them as
the bridge between different apps and groups.

James