[DAS2] single source document format

Andrew Dalke dalke at dalkescientific.com
Sun Jan 15 23:31:02 UTC 2006


I've been working on updating the spec.  So far working on the
sources document(s).

We have 2 or 3 document formats, depending on how you count.

  - the top-level URL for a server returns a 'sources' request,
       which can return 0 or more sources (each with 0 or more versions)

  - the top-level URL for a given organism/database returns a
       'source' request which returns 1 source (each with 0 or more 
version)

  - a 'versioned source' request, which returns more details about
      the given version of the source (one version)

Currently the first 2 return the same document type, which is
similar to but different from the 3rd.

** I propose that all three return a document of the same content
type, instead of returning 2 different document types. **

"How are they currently different?" you ask.

The versioned source request returns a bit more information
than the normal source request.  It returns the 'NAMESPACE' fields,
which point out that 'features' are available at XYZ and 'types'
available at ABC.  It also (now) has a key/value table for
metadata properties that Andreas (Prlic) wanted.


NOTE: there would still be three URLs involved.  The only difference
in this proposal is that all of them return a document of the
content-type 'application/x-das-sources+xml' and there is no need
to drill down to the 3rd level URL if given the results from
the 1st level URL.



Advantages:
   - a single request to a data server (including to the metadata
       server) to get the main information about a given data source
       and version.  That is, no need to go to the versioned source to
       ask it for the place to get information about the 'types'
       or 'properties' URL.

   - Andreas pointed out that he needs a bit more information in
       the SOURCES for the registry server.  This is also information
       that should be available from the database/version itself.
       I would prefer that they have the same format and structure,
       for better code reuse.

(BTW, another name for 'registry server' might be 'DAS aggregator'.)

Disadvantages:
    - the registry server must return more data (but there isn't
       that much)

    - there's a better chance of registry server data invalidation
       (though those fields don't change much)


As an alternative, they could have the same format but the
registry server returns only a subset of the data, so a
client can query the registry server to find the versioned
source URL then query that URL to get the complete data set.

					Andrew
					dalke at dalkescientific.com




More information about the DAS2 mailing list