Re: [ubl-ndrsc] More comments on Tag Structure paper

[Dave Carlson <dcarlson@ontogenics.com>]

Also, Mike, if I understand another one of your comments, the "Party"
and "Measure" are actually datatype information that is appended to
the tag name.  This seems really wrong to me.  The datatype of patient
and weight are defined in the schema.  So, why not:

<patient>
  <weight>
  </weight>
</patient>

And the schema would have:

<xsd:element name="patient" type="Party"/>
<xsd:element name="weight" type="Measure"/>

where the Party and Measure are presumably defined as xsd:simpleType
datatypes.


----- Original Message -----
From: "Mike Rawlins" <mike@rawlinsecconsulting.com>
To: "Eduardo Gutentag" <eduardo.gutentag@sun.com>
Cc: <ubl-ndrsc@lists.oasis-open.org>
Sent: Monday, November 26, 2001 5:20 PM
Subject: Re: [ubl-ndrsc] More comments on Tag Structure paper


To be completely consistent with my example, I think what Eduardo
means is he prefers:

<patientParty>
                <weightMeasure>
                </weightMeasure>
</patientParty>

to

<patientParty>
                <patientPartyweightMeasure>
                </patientPartyweightMeasure>
</patientParty>

At any rate, I think he agrees with my suggestion of omitting the
object class name.

Eduardo Gutentag wrote:

> This brings up something I queried Eve about last week and she
suggested I bring
> this up in this forum. Unfortunately I have not had time until now.
>
> Why are we, in an XML-centric activity, even considering things like
> <patientPartyWeightMeasure>? I can see the need for this in
environments
> where the "markup" is flat, but in XML? Why not deal simply with
> <Patient>
>         <Party>
>                 <Weight>
>                         <Measure>
>                         </Measure>
>                 </Weight>
>         </Party>
> </Patient>
>
> where the hierarchy gives you all you need in terms of semantics,
reusability, etc.?
>
> How can one reuse <patientPartyWeightMeasure>? No way.
>
> If there is a need to map to some other method, doing it by
accumulating
> parents of a given element should be quite trivial.
>
> Can someone throw some light on this? Is there some lore behind this
that
> I'm ignorant of? Seems to me the only reason this is being
contemplated is
> that's the way it's always been done - and that's not good enough,
IMHO.
>
> Thanks
>
> Mike Rawlins wrote:
> >
> > I basically favor Mark's option 1, but with some variations,
> > completions, and comments.
> >
> > Option 1 is incomplete in that it specifies how element (etc.)
names are
> > to be constructed using the assigned representation term, but it
doesn't
> > say anything about names for the component parts.  For example, we
might
> > have a basic information entity of patientPartyWeightMeasure.  The
> > approach only gives us a tag for the "content" part of the
information,
> > but not for the unit of measure associated with every item that
has a
> > representation type of measure.  Assuming that the "content" is
the
> > element content and other components of the representation term
are
> > expressed as XML attributes (an approach we might want to
consider, but
> > that's a different discussion), do we say:
> >
> > <patientPartyWeightMeasure
uom="KGM">140</patientPartyWeightMeasure>
> >
> > or
> >
> > <patientPartyWeightMeasure
> > unitOfMeasure="KGM">140</patientPartyWeightMeasure>
> >
> > or
> >
> > <patientPartyWeightMeasure
xxx="KGM">140</patientPartyWeightMeasure>
> >
> > where xxx is something else as yet undefined?
> >
> > I also favor an approach where the "Object Class" portion of the
name is
> > omitted for the children.  For example, instead of having:
> >
> > <patientPartyDetails>
> >     <patientPartyNameText> ...
> >
> > we have
> >
> > <patientPartyDetails>
> >     <nameText>
> >
> > The latter approach is much less verbose and allows an XML schema
> > representation where the CC might be a type (like party) and all
we need
> > to do is declare or extend the type for the BIEs (like
patientParty).
> > Otherwise, we have to declare each and every <mumblePartyNameText>
in
> > the schemas and have practically *no* reuse at the XML syntax
level.
> >
> > Finally, though I don't have as strong preferences on this and
can't
> > provide as good a rationale, I favor dropping the representation
term
> > part of the name unless it is needed to avoid duplication.  Saying
> > "partyNameText" instead of "partyName" seems overly verbose and
adds
> > little value if the information is understood in context.  A
developer
> > can always look at the element or type definition if the data type
is
> > not clearly understood from a knowledge of the business data.
> >
> > --
> > Michael C. Rawlins, Rawlins EC Consulting
> > www.rawlinsecconsulting.com
> >
> > ----------------------------------------------------------------
> > To subscribe or unsubscribe from this elist use the subscription
> > manager: <http://lists.oasis-open.org/ob/adm.pl>
>
> --
> Eduardo Gutentag               |         e-mail:
eduardo.gutentag@Sun.COM
> XML Technology Center          |         Phone:  (510) 986-3651
x73651
> Sun Microsystems Inc.          |

--
Michael C. Rawlins, Rawlins EC Consulting
www.rawlinsecconsulting.com



----------------------------------------------------------------
To subscribe or unsubscribe from this elist use the subscription
manager: <http://lists.oasis-open.org/ob/adm.pl>