[ckan-discuss] Guidance for use of Author and Maintainer fields

Rufus Pollock rufus.pollock at okfn.org
Fri Jul 30 10:18:51 BST 2010

On 28 July 2010 10:51, William Waites <william.waites at okfn.org> wrote:
> On 10-07-28 02:43, Glen Barnes wrote:
>> The OSM method seems to work quite well. Essentially there are
>> proposals for tags that can be discussed and then ratified as being
>> official. We don't have to go down the whole voting route just yet but
>> why don't we do this:
> This is getting perilously close to the reason for RDF...

It is and we've been aware of this for a while -- if only RDF stores
(or even nosql stores) had been mature when we started 4+ years ago

> I would suggest that rather than the willy-nilly approach that
> OSM uses (horse=yes) we try as far as possible to use tags
> that can be mapped easily to RDF predicates.
> Proposal #1: Tags SHOULD be of the form "prefix:tag" where
> prefix is the usual well-known prefix that would be used in
> an RDF/N3 serialisation.
> Proposal #2: Have a special prefix, "prefix" to explicitly define
> prefixes to be used elsewhere in the record. So you might have,
>    prefix:dc=http://purl.org/dc/terms
>    dc:description=Some description

I think we should try and find a balance between people being free to
just "go ahead" and a decent structure so we don't end up with lots of
random "extension fields" (or tags)

> #2 might prove a bit awkward in practice though.
> Is it time to start looking at ripping out the SQL back-end and
> replacing it with an RDF store? Is there anyone that is
> interested in helping to do this? If we could keep as much of
> the user-interface code as possible it shouldn't be too hard.

Yes to both: is it time and would we be interested :)

It definitely is time to think about this but it would clearly have to
go on in parallel. This should probably become a separate thread ...


More information about the ckan-discuss mailing list