[eml-dev] [Bug 3480] - duplicate distribution types (in resource.xsd, and physical.xsd)

bugzilla-daemon at ecoinformatics.org bugzilla-daemon at ecoinformatics.org
Fri Aug 29 15:14:10 PDT 2008


http://bugzilla.ecoinformatics.org/show_bug.cgi?id=3480





------- Comment #2 from mob at icess.ucsb.edu  2008-08-29 15:14 -------
Summarizing comments on eml-dev:
The distribution element in the resource group is generally intended to be
informational, although sometimes it has been used for download. The
distribution element in the entity group is generally intended to be machine
readable, and used for download.

looking closely at the 2 named Types and their uses, this is what I think
should happen:

First, in eml-resource.xsd:
1. keep named type, DistributionType in the resource group.
2. make the element offline/mediumName required, per bug #1154
3. make sure that the documentation on the res:distribution element reflects
the recommended or allowed uses, especially wrt the use of  function="download"
or "information". 


Second: in eml-physical.xsd
1. derive a second type PhysicalDistributionType from res:DistributionType.
Currently, it is a copy of res:DistributionType (with some additional
documentation), and gets some of it's simpleTypes from the eml-resource schema.
2. The derived Type can have additional elements (like <access>, see bug #1132
for code)
3. update the documentation on the physical/distribution element to reflect the
recommended use


And third: eml-software.xsd:
this schema actually has two <distribution> elements: the first in its resource
group, and the second as a child of <implementation>.
1. The first is untouched, obviously.
2. the second (implementation/distribution) could be retyped to the new
PhysicalDistributionType. This makes the <access> tree available to
implementation
3. clarify the appinfo on the implementation/distribution, since currently,
this is a duplicate of the resource tag.


I tried to make schema language reset the default attribute value for the
extended Type so that a particular use could be reflected in the model (ie,
function="information" or function="download". Maybe this is not necessary, or
maybe someone else has done this and can advise? As is, the default of
"download" implies the most stringent content, but can be relaxed if authors
need to.


More information about the Eml-dev mailing list