[Go-essp-tech] DOI target page CMIP5

[Bryan Lawrence]

hi Karl

Most of these issues were discussed in the doi paper or in the 
discussion around it.

1) DOIs are to be granted per simulation (in metafor terms), that is to 
aggregations of realm publication units, themselves aggregations of 
atomic datasets. (Hence, o(1000) DOIs for CMIP5)
 - the URLS and identifiers we have for those smaller units are then 
analogous to chapters and pages.

2) The issue of "replacement etc" is why we have a comprehensive 
procedure to minimise the necessity for replacing things, but the entire 
machinery is predicated on the fact that indeed there will be versions.

3) DOIs will always point to pages which point at the actual aggregation 
at the time of DOI allocation, but the page can be marked as superceded.

Cheers
Bryan

> Hi all,
> 
> Has it been decided how large a chunk of data is going to be assigned
> a DOI.  Will they be assigned at the DRS atomic dataset level?  Or
> the DRS "publication-level dataset" level? or what?   I guess the
> tradeoff is:
> 
> Larger chunks require fewer DOI's
> Larger chunks mean that some files will be associated with multiple
> DOI's (when a single file in the chunk gets replaced, the chunk will
> have to be assigned a new DOI, so all the other files will now be
> associated with 2 DOI's).
> 
> We will have to provide a service so a user knows what data is
> associated with each DOI.  If a given subset of a chunk of data is
> associated with multiple DOI's, a reference to either DOI should
> point to the same (identical) subset of the data.
> 
> Is this confusing?
> 
> best regards,
> Karl
> 
> On 11/16/10 5:22 AM, Martina Stockhause wrote:
> >    Hallo Bryan, dear all,
> > 
> > we started to set up the mirror pages of your DOI target page. It
> > is very rough, since we plan to use your stylesheet and the
> > link(s) to the data at the three ESGF locations are not
> > implemented, yet. We filled the page with our CERA test data.
> > 
> > We suggest to put the summary on the page, because otherwise there
> > is no immediate information available. The summary uses sentences
> > from the CMIP5 page and the experiment description of the
> > questionnaire (atomfeed experiment).
> > 
> > What is missing:
> > - construction of the link into the CIM repository to the
> > simulationRun document to replace the cirrus-link.
> > - datanode TDS root catalogue and file server root access for the
> > data access link.
> > - style sheet of your primary DOI target page.
> > 
> > To construct a link to the data at BADC and PCMDI we need the TDS
> > root and the file server root addresses. We plan to provide lists
> > of chunks for the given DOI and a list of ESG dataset links.
> > 
> > E.g. DKRZ:
> > tds root (replicated data):
> > http://BLOCKEDbmbf-ipcc-ar5.dkrz.de/thredds/esgcet/1/ fileserver
> > root: http://BLOCKEDbmbf-ipcc-ar5.dkrz.de/thredds/fileServer/new/
> > 
> > Please add your information for BADC and PCMDI at
> > http://BLOCKEDesgf.org/wiki/Cmip5Status .
> > Just to make sure: You will use one root for all replicated data?
> > If not, we cannot provide the link list but only the DRS names
> > (TDS file_id + version + variable in case of chunks and TDS
> > dataset_id in case of ESG datasets).
> > 
> > What do you think? Any suggestions for improvements?
> > Any suggestions what the DOI service should provide for the users /
> > the portals? The chunk list for a DRS experiment will include
> > hundreds of links, which is not very convenient for the user.
> > 
> > Best wishes,
> > Martina

--
Bryan Lawrence
Director of Environmental Archival and Associated Research
(NCAS/British Atmospheric Data Centre and NCEO/NERC NEODC)
STFC, Rutherford Appleton Laboratory
Phone +44 1235 445012; Fax ... 5848; 
Web: home.badc.rl.ac.uk/lawrence