Re: Proposal to modify Classifications in ebRIM

[Farrukh Najmi <Farrukh.Najmi@Sun.COM>]

Martin,

I just saw that you had mentioned returning "any fragment" from document tree in
your  use case. This is not supported at the moment. I neglected to mention that the

spec need to be tightened to indicate that the XPATH expression must yield a
ClassificationNode.

Returning fragments is something we have not focused on at all in V1.0. How
important is this use case? If it is important then we should have an ability to
specify an XSLT filter and return an arbitray XML fragment.

IMO returning fragments is a separate use case from Classification schem support use
cases as it cuts across many other non-scheme related use cases.

Thoughts?


Farrukh Najmi wrote:

> Martin,
>
> Glad we can meet use case #1. For use case #2 see my comments at the bottom.
>
> Martin Bryan wrote:
>
> > >To go the other direction it works as well. One can issue a
> > GetClassificationTreeRequest and give it  ClassificationNode reference to
> > any
> > node in any scheme, and a integer depth parameter as input. The resulting
> > GetClassificationTreeResponse will have the complete linked tree for the
> > requested sub-tree in the scheme scheme as a single XML document when the
> > depth
> > specified is negative. Othewise it return descendent nodes upto specified
> > level
> > of depth. If the specified node is a root of a scheme then the entire scheme
> > can
> > be retrieved as well.
> >
> > >Does that meet the first use case as you envision it?
> >
> > Yes
> >
> > >
> > > getPath only returns a single node. Note that I set "set of nodes". Using
> > a
> > > range statement I should be able to retrieve any fragment of an existing
> > > classification scheme that is relevant.
> >
> > >The above can currently be supported with SQL Query facility in OASIS ebRS
> > V1.0
> > with a query like below:
> >
> > >SELECT FROM ClassificationNode WHERE path = ‘/Geography/Asia/*’
> >
> > This seems to offer no functionality over the above as Asia is a specific
> > node, and * seems to indicate all entries one level down from this. I was
> > thinking more along the lines of Europe[5,15] to select a specific set of
> > nodes, such as those forming the EU.
> >
> > Martin
>
> Your example for use case 2 brings up a larger issue that we do not support the
> ability to use the same ClassificationNode in multiple schemes. For example UK
> should be a child of Europe as well as a child of EU. This can be done by
> classifying classification nodes but it is awkward.
>
> That said....
>
> The semantic of the spec are that path is an XPATH expression. While it may not
> be clearly stated in the spec, the intent of moving to an XPATH expression was
> to be able to do any valid XPATH expression that would operate over the document
> tree representing all the ClassificationNodes that match all other opredicates
> of the query. Thus your use case could be handled by the following modified
> query:
>
> SELECT FROM ClassificationNode WHERE path = ‘/Geography/Europe[5,15]’
>
> Note however that relying on particular ordering in the document tree for EU
> membership is risky.
>
> Does the above satisfy the use case?
>
> --
> Regards,
> Farrukh

--
Regards,
Farrukh

begin:vcard 
n:Najmi;Farrukh
tel;work:781-442-0703
x-mozilla-html:FALSE
url:www.sun.com
org:Sun Microsystems;Java Software
adr:;;1 Network Dr. MS BUR02-302;Burlington;MA;01803-0902;USA
version:2.1
email;internet:najmi@east.sun.com
fn:Farrukh Najmi
end:vcard