DIDs in DAL [was: SODA, half-client...]

[Markus Demleitner]

Hi François,

On Fri, Apr 15, 2016 at 07:24:48PM +0200, François Bonnarel wrote:
>    -   In your obstap califa collection, the obs_creator_did is empty. I
> guess the califa creator is Calar Alto observatory. Does that mean this
> observatory has no will to define a creator id for its products? Or maybe
> it's delayed for whatever reason?

In this case, the data products came from the CALIFA collaboration
rather than Calar Alto directly.  I certainly could have talked to
them about IVOA creator DIDs, but I didn't. Should I have?  Since I'm
not aware of anything actually making use of creator DIDs in the
context of Obscore, I figured it's better not to complicate matters
for them.

Does anyone else (other than HiPS, for which these actually have a
function) try to make data providers assign IVOA-compliant creator
DIDs (i.e., <project IVOID>?<local part>)?  If so, what do you tell
them as to why they should do this?

>     - 2nd point  : ob_publisher_did for califa cube in the Obscore table
> follow the template:
> **ivo://org.gavo.dc/getproduct#califa/datadr2/?????.????.??????.fits
>    But this publisher_did is not used to call the associated {links}
> resource response from DataLink. Instead is used something with the
> following template :
> ivo://org.gavo.dc/~?califa/datadr2/?????.?????.?????.fits
> 
>    It is perfectly allowed to call "DataLink" with an internal  ID or an ID
> different than the publisher_did.
>    But here the iD used for DataLink also look like a registered
> ob_puublisher_did. Only it is different.
>    So my question: do you have two different publisher_did for the same
> product ? And if yes why ?

That's a consequence of the migration from the old PubDIDs (abusing
fragment identifiers) to the new ones (employing query parts for the
local part).  See Norman's Note on URI fragments
(http://ivoa.net/documents/Notes/URIFragments/index.html) on why
we're doing this.

During the migration, some of our services temporarily understand
both the legacy old-style and the new-style PubDIDs.  For CALIFA, the
database still has the old-style ones since we've not re-imported the
data.

In case you're curious: ivo://org.gavo.dc/~ is indeed registred (try
it on http://dc.g-vo.org/WIRR :-).  It's our data delivery service
giving access to all published data products (and doing, e.g.,
authentication).  It's planned that it also grows a datalink face,
but the assignation of the datalink service responsible for a given
data product to the local ids is something that needs to be done
manually (if there's a datalink service at all), and so that's a bit
on the back burner.

Cheers,

         Markus