[seek-dev] anomolies in resultset.xsd

Kevin Ruland kruland at ku.edu
Wed Nov 16 17:29:57 PST 2005

Hi all,

In my effort to fix the resultset de-/serialization problem I've
uncovered some wierd things in the resultset.xsd.  I will present each
with my proposed solutions.

1) //record/returnField has an element named <content>.  So a returned
file would look like this:  <returnField
id="f9"><content>XX</content></returnField>.  I don't really understand
why this element exists since the content of the field should really be
the text in the returnField element itself.  I recomment we remove this
element because it just makes the resultset documents much larger than
necessary (my guess is around 3-5% larger).

2) None of the elements other than the root are in a namespace.  They
are all in the default namespace.  This is because of the extensive use
of local element definitions.  It can be "fixed" by setting the
elementFormDefault attribute in schema to "qualified".  I suggest we
make this change since not using a namespace is so five minutes ago.

3)  The record element has two required attributes lastModifiedTime and
creationTime.  These elements are currently not being pulled & populated
by the digir server.  Since these are not nil-able, and required there
is no way to return "unknown".  I suggest we make them optional. 
Additionally, since I have a lot of work to do, I suggest we ignore the
fact that Digir isn't producing this data now and log it as a bug to be
fixed later.

Overall, I'd like to make any and all changes in the axis branch and let
the globus branch wither.  And I'd like to do it sooner rather than
later.  If there are no objections I will make these changes Thursday in
the branch.


More information about the Seek-dev mailing list