Datalink Feedback VI: Semantics

[Patrick Dowler]

We have recently updated our datalink service to try to shake out all 
the issues we have discusued here.  On the topic of "full dataset 
retrieval" and service_def, we just opted to leave service_def 
empty/null (as in the last WD), which means this URL can be used as is.

On our portal, the download application calls datalink and in normal 
mode just picks out all the non-service urls (eg service_def = null) and 
uses them.

The datalink response includes real service links (service_def = 
something and there is an associated resource with that ID), but we do 
not yet have the intermediate case: a link which is not part of the 
dataset retrieval but is an opaque url, such a a canned link to a 
graphical preview image. We have such things, and if I was to add that I 
would want to be able to differentiate it from "dataset retrieval"...

I would expect the semantics column to provide the solution to it.

So, I still think on the datalink service side a null service_def and 
null or special ("self") semantics is correct for "dataset retrieval".


Pat

On 14/04/14 06:02 AM, François Bonnarel wrote:
> As far as the "full retrieval" is concerned (self) I don't think it
> belongs to the dataset. I think "dataset reterieval" is some kind of
> service and should be described via the "service-def" field.

IMO, A service_def would really obscure the fact that this is a plain 
old download.

-- 

Patrick Dowler
Canadian Astronomy Data Centre
National Research Council Canada
5071 West Saanich Road
Victoria, BC V9E 2E7

250-363-0044 (office) 250-363-0045 (fax)