[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: documentation

Hi Ethan

On Thursday 05 January 2006 17:55, Ethan Davis wrote:
> Bjørge Solli wrote:
> >I'm using thredds 3.4 on tomcat 5.0.30.
> >
> >My content/thredds/catalog.xml contains:
> ><catalogRef xlink:title="North Atlantic Class 1 - Best estimate and
> > bulletins" xlink:href="dodsC/mersea-ip-class1-nat.xml">
> >     <documentation type="summary">
> >                   DODS Data Prepared for the MERSEA IP EU Project.
> >     </documentation>
> >     <documentation type="simple"
> >                   xlink:href="http://www.mersea.eu.org";
> >                   xlink:title="MERSEA Project Web page">
> >     </documentation>
> ></catalogRef>
> >
> >Where is the documentation shown?
> It really depends on what THREDDS client you are using. It sounds like
> you are viewing the catalogs in a browser in which case the TDS XML to
> HTML code can be thought of as the THREDDS client. In the TDS HTML
> catalogs, the documentation is only seen on datasets that have IDs (and
> are not catalogRefs, as those are links to the next level of catalog.
> The NetCDF-java ToolsUI client displays the metadata for whatever
> dataset is currently selected.

Thanks. This helps. I am myself not a user of the data, I'm just aiding the 
researchers, so I did not think of this.

> >Wouldn't it be better to have a <metadata inherited="true">...</metadata>
> >inside <catalogRef> and let all the datasets in
> >dodsC/mersea-ip-class1-nat.xml inherit them? Another way that would work
> > for me was to have it like this:
> ><dataset "container"> <!-- no ID on this -->
> ><metadata inherited="true">...</metadata>
> ><catalogRef ... />
> ></dataset>
> >And then all the datasets in the catalogRef also inherit the metadata.
> Unless the catalogRef and the referenced catalog are generated from
> datasetScan elements, the metadata in your top catalog will not be
> visible across the catalogRef link in the referenced catalog even if the
> metadata is declared inherited. Since each catalog can be referenced
> seperately, a catalog needs to stand alone. It can reference other
> catalogs and URLs but it can not depend on an upstream link that the
> user/client may not have followed. Here's how the spec puts it, "[t]he
> referenced catalog is not textually substituted into the containing
> catalog, but remains a self-contained object".
> Of course, you can put the metadata in the referenced catalog by hand so
> that it is inherited by all the contained datasets. Or you can have the
> catalog(s) generated from datasetScan elements and all the metadata for
> the datasetScan will be automatically included in the generated catalogs.

My catalogs are big and generated by a script. I'm looking into the 
possibility of using catalogScan, but at the moment this seems not practical. 
I will just have to make the script add the metadata as well.

> >I use catalogRef to avoid having to process all my large datasets unless
> >specifically asked for, but they mainly have pretty much the same
> ><authority>, <dataType>, <dataFormat>, <documentation>, <creator> and
> ><publisher>.
> Are you using datasetScan elements for automatic catalog generation in
> your content/thredds/catalog.xml? Or only at lower levels of you catalog
> tree. Using datasetScan elements would do a lot of this for you and the
> catalogs don't get generated until requested.
> We use catalogRef elements in our TDS configuration to split things
> apart that have very different metadata. And just to keep the pieces we
> need to maintain on a reasonable scale. The problem with using
> catalogRefs is that it often requires a bit of metadata duplication.
> Hope this helps. Let me know if you have further questions.

Thanks. It helped.

Bjørge Solli - Office:+47 55205847 cellph.:+47 91614343
Nansen Environmental and Remote Sensing Center - Bergen, Norway
http://www.nersc.no Reception: +47 55205800
Dept.: Mohn-Sverdrup Center for Global Ocean Studies 
       and Operational Oceanography

NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.