new characterization schema: STC Concerns and Proposal
Arnold Rots
arots at head.cfa.harvard.edu
Thu Sep 28 12:09:49 PDT 2006
Dear Francois,
AstroCoordSys, AstroCoords, and AstroCoordArea are indeed not
low-level, but I thought that you actually did use some lower level
elements.
However, the main point I was making is to separate the model from the
XML implementation. I fully agree on the complementarity of the
Characterisation and STC models, but suggested that both models can be
adequately represented in XML through a single implementation: the
ObsDataLocation element as defined in the STC schema.
What I noticed in your and Anita's talks was that in the XML
implementation that you were proposing there are still elements that
are not properly connected (as Igor also pointed out). Fixing that is
not quite as trivial as Jonathan suggested. Instead, I would submit
that all the functionality is already provided in ObsDataLocation.
That was not the case in the past, but with the reorganization of the
derivation tree in the schema, it is now complete for all coordinate
axes. And that means that this element has all the functionality to
serve as an implementation of the STC model as well as of the
Characterisation model.
There are various ways to link a model to an implementation. I am not
sure utypes are the best way, but it works and is actually simple:
all you need is a utype mapping table that provides the translation
between char utypes and stc utypes.
Cheers,
- Arnold
Francois Bonnarel wrote:
[ Charset ISO-8859-1 unsupported, converting... ]
> Dear Arnold, dear all
>
> I think we touch here the very heart of the STC/Characterisation
> complementarity.
> As I tried to explain in my talk in Moscow, Characterisation is there
> to describe the observations and the data (while STC dezcribes Coordinates
> and Areas/Intervals in rigourously defined Coordiante systems) .
> Char semantics and meaning is not only in the structure but also in the
> attribute names (or utypes) organized in a matrix of properties, axes and
> Levels. All this stufF has beeen extensivly discussed in the sub-group
> since 3 years and seem to make sense for data providers and archive managers.
> At some point, char need to use coordinate descriptions , and to define areas
> in the sky or on parameter axes and we find a rigourous description with
> STC. I never thought that AStroCoordSys , AstroCoords and AstroCoordArea
> which we are using are low-level elements. According to Jonathan (last
> utype document) and I guess to you it is actually the heart of STC data model
>
> OBsDataLocation and ObservationLocation on the contrary are conveniance
> high level structures more or less at the level of charac but which cannot
> replace it. If some VO applications(VO Event) are happy to use them I think t
> hat's fine but I cannot understand how a data model which you say has not
> to be changed in your previous mail (characterization) can be implemented using
> the sole schema which contain none of the utypes and structuration coming
> from Characterization.
>
> In my next mail I will tell that it is even not necessary because STC
> AstroCoords, AstroCoordSys, and AstroCoordArea contain everything which
> allow rigourous definition and linkage.
>
> Cheers
> Fran?ois
>
> <From owner-dm at eso.org Tue Sep 26 20:59:45 2006
> <X-UID: 389
> <
> <The question about encapsulating elements really is: at what level?
> <ObsDataLocation obviously can be regarded as a public class, but some
> <of the low-level elements are really private classes since they lose
> <some of their functionality when taken out of context.
> <I guess I was suggesting to let ObsDataLocation be the XML schema
> <implementation of Characterisation, in which case the indirection is
> <not required anymore.
> <But see what you can work out.
> <
> < - Arnold
> <
> <louys at alinda.u-strasbg.fr wrote:
> <[ Charset ISO-8859-1 unsupported, converting... ]
> <>
> <>
> <> Dear Arnold,
> <>
> <> My first impression is that your suggestion would impose an overhead in
> <> the characterisation strucure , because each characterisation item , at
> <> the 1-2 levels will have to point to the ObservationLocation structure
> <> which means one added level of indirection.
> <> I would really prefer to use the STC structures in an object oriented
> <> way , that is encapsulating STC elements.
> <>
> <> we 'll work out examples on instance documents, to check for the
> <> feasibility of your approach in our context.
> <>
> <> cheers, Mireille
> <>
> <> Quoting Arnold Rots <arots at head.cfa.harvard.edu>:
> <>
> <> > When I asked yesterday why Characterisation wasn't using higher-level
> <> > structures from STC, I was actually serious and even more so after
> <> > Anita's presentation and some of the concerns that Igor voiced.
> <> > I realized that the Characterisation schema is going through the same
> <> > issues now, that I went through in the development of the STC schema,
> <> > in terms of required axes, elements, and references. It seems
> <> > unnecessary to have to rehash all of that again.
> <> > In particular, by rebuilding the STC-related elements from lower level
> <> > STC components the integrity of the coordinate metadata in the related
> <> > axes is lost and it will be a major effort to regain it.
> <> > The thing is that all of the axis-related stuff in Characterisation is
> <> > contained in the ObsDataLocation element in STC, with the exception of
> <> > sensitivity. It would be a rather simple thing to add the sensitivity
> <> > concept to stc:AstroCoordArea. When that is done you would have a
> <> > complete implementation of the axis-related characterisation in
> <> > stc:ObsDataLocation, complete with all the proper structure and added
> <> > value in a fully implemented referencing mechanism.
> <> > It is easy to see that Anita's example would be completely represented
> <> > in this - I'll provide an example of it next week.
> <> >
> <> > I am not proposing any change in the Characterisation _model_, only in
> <> > the XML schema implementation by simply plugging in ObsDataLocation
> <> > which is a ready, complete, and tested element.
> <> > Note that this same ObsDataLocation element is also used in the
> <> > VOEvent schema.
> <> >
> <> > I know that the organization of the information is different - the
> <> > matrix is transposed, so to speak. But that was done to facilitate an
> <> > easier implementation of the related coordinates. And it is only a
> <> > matter of code: it should not be harder to present the information in
> <> > a transposed view.
> <> >
> <> > And I don't think it would delay the process, either. As I said, the
> <> > model (and hence the WD) is not affected, while the schema development
> <> > will be easier since all coordinate concerns are taken care of. I
> <> > promise to add the sensitivity components right away to STC since it
> <> > does not seem like that will be difficult since I should be able to
> <> > take it from the current Characterisation schema.
> <> >
> <> > Cheers,
> <> >
> <> > - Arnold
> <> >
> <> > Francois Bonnarel wrote:
> <> > [ Charset ISO-8859-1 unsupported, converting... ]
> <> >>
> <> >> Hi all,
> <> >> Version 0.95 of the characterization xml schema (axis first) is available
> <> >> following this link :
> <> >> http://Model/Characterisation/schema/characterisation.0.95.xsd
> <> >> It has been properly validated, importing stc1.3 schema and using
> <> >> xmlspy. So it could be possible to use it to develop validated examples.
> <> >>
> <> >> Best regards
> <> >> Fran?ois
> <> >>
> <> >> =====================================================================
> <> >> Francois Bonnarel Observatoire Astronomique de Strasbourg
> <> >> CDS (Centre de donnees 11, rue de l'Universite
> <> >> astronomiques de Strasbourg) F--67000 Strasbourg (France)
> <> >>
> <> >> Tel: +33-(0)3 90 24 24 11 WWW:
> <> >> http://cdsweb.u-strasbg.fr/people/fb.html
> <> >> Fax: +33-(0)3 90 24 24 25 E-mail: bonnarel at astro.u-strasbg.fr
> <> >> ---------------------------------------------------------------------
> <> >>
> <> > --------------------------------------------------------------------------
> <> > Arnold H. Rots Chandra X-ray Science Center
> <> > Smithsonian Astrophysical Observatory tel: +1 617 496 7701
> <> > 60 Garden Street, MS 67 fax: +1 617 495 7356
> <> > Cambridge, MA 02138 arots at head.cfa.harvard.edu
> <> > USA http://hea-www.harvard.edu/~arots/
> <> > --------------------------------------------------------------------------
> <> >
> <>
> <>
> <>
> <>
> <--------------------------------------------------------------------------
> <Arnold H. Rots Chandra X-ray Science Center
> <Smithsonian Astrophysical Observatory tel: +1 617 496 7701
> <60 Garden Street, MS 67 fax: +1 617 495 7356
> <Cambridge, MA 02138 arots at head.cfa.harvard.edu
> <USA http://hea-www.harvard.edu/~arots/
> <--------------------------------------------------------------------------
> <
>
> =====================================================================
> Francois Bonnarel Observatoire Astronomique de Strasbourg
> CDS (Centre de donnees 11, rue de l'Universite
> astronomiques de Strasbourg) F--67000 Strasbourg (France)
>
> Tel: +33-(0)3 90 24 24 11 WWW: http://cdsweb.u-strasbg.fr/people/fb.html
> Fax: +33-(0)3 90 24 24 25 E-mail: bonnarel at astro.u-strasbg.fr
> ---------------------------------------------------------------------
>
--------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701
60 Garden Street, MS 67 fax: +1 617 495 7356
Cambridge, MA 02138 arots at head.cfa.harvard.edu
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------
More information about the dm
mailing list