[okfn-help] Multiple download URLs for CKAN

[Jonathan Gray]

(This is perhaps a bit tangential - but I think relevant to our longer
term plan in this area.)

I remember a while ago Rufus mentioned to me the idea of using scripts
(instead of, or as well as URL) to grab material associated with a
package. As well as getting material, I guess these would assist in
automated backups or mirrors. Might the multiple download URLs be used
in such scripts in future?

Re: distributions/resources nomenclature - is this just nomenclature,
or does how CKAN will function somehow depend on what we decide here?
If its just names - what about 'bulk' and 'part' instead of
distribution/resource (which is perhaps less obvious).

I guess another thing to think about is - when there are multiple
versions of a package (e.g. if someone took CSV file and converted to
RDF and wanted to link to it on CKAN) - are we currently planning to
put this on the main package page, or to split into different pages? I
guess my feeling is we want to put on same page until it changes
significantly enough to deserve its own page. Another thing to think
about is forking or packages and dependencies between packages. Do we
have a longer term plan here?

Also - what about textual descriptions in different languages
associated with the same package. I guess most of the metadata will be
the same? This will become something to think about when we have
German version of CKAN.

(Sorry - I've strayed off topic of download URLs, feel free to
disregard irrelevant bits!)

Jonathan

On Tue, Nov 17, 2009 at 10:39 AM, David Read <david.read at okfn.org> wrote:
> We're specifying a multiple download URLs feature for CKAN packages
> and have some questions. See ticket:
>
> http://knowledgeforge.net/ckan/trac/ticket/189
>
> Question 1:
>
> We're discussing multiple download urls for a package. Here's the
> ticket: http://knowledgeforge.net/ckan/trac/ticket/189
>
> Distributions vs resources
>
> Distribution is: the whole of a package's payload (in one file or url/api)
> Resource: is any single file associated with a package
>
> e.g. with wikipedia may have a full dump in SQL, HTML or JSON format.
> Each is a 'Distribution' - the complete thing but in a different
> format. But they may also have separate files containing a part of the
> whole, e.g. A-M.zip and N-Z.zip which we term 'Resources'.
>
> At the moment we're focusing on supporting Distributions. Our concern
> is that this will be quickly abused for Resources. Is this likely?
> And, if so, should wee just bite the bullet and get rid of the
> distinction right now?
>
> Question 2:
>
> We think a Distribution (or a Resource) should have a URL, a format
> and a description. Does this sound right? Also, should we rename
> 'description' to 'title' to imply this should be only a short summary.
>
> David
>
> _______________________________________________
> okfn-help mailing list
> okfn-help at lists.okfn.org
> http://lists.okfn.org/mailman/listinfo/okfn-help
>



-- 
Jonathan Gray

Community Coordinator
The Open Knowledge Foundation
http://www.okfn.org