Model updates..

CresitelloDittmar, Mark mdittmar at cfa.harvard.edu
Tue Apr 3 23:53:06 CEST 2018


All,

I had a lengthy mail that I sent out a few days ago.. has bounced back.

Rather than re-try.. here is a summary.
  + DatasetMetadata, Cube model documents (PDF) and vo-dml products
(HTML/XML) are updated in volute.
  + The 'stc' models (coords, meas, trans)
      * The 'docs' section in volute has diagrams and documentation for
Arnold's
         last drop of the model.  There is a good bit of migration from
this base on
         pre-Santiago example generation and recent work with the TDIG
example set.
       * VO-DML/XML and HTML are updated in volute to current.
         NOTE: in the coming week (this one), I'll be making a pass to
remove the
         domain specific 'general' coordinates.  With the specialized
shortcuts, these
         seem less important.
  + Example files include
      * STC snippets in the STC volute space (to be expanded)
      * Cube - sparse cube (event list)
      * Cube - 2D image
      * Cube - 4D image
      * Simple Time series as cube
      * prototype serialization of Vizier time series as cube
      ** Examples and demos in Omar's notebook

All the above is documented in the respective wiki.

Going forward:
  In generating the cube examples, everything looks good to me, except for
describing virtual data.  ( eg: eqpos(ra,dec) = TProject( sky(x,y) ) or
image axes)
   The previous modeling had FrameTransform relating two Frames together by
the Transform operation.  However, there is nothing in the file or model
which would inform a client that this particular column is generated by
transforming another column.  Or that the image has sky and world axes.  Or
the image value is scaled.
    I have outlined a change which I think will resolve this problem.
Rather than the generic "transform connects 2 frames", I'd like to define
specific VirtualObject types which relate a source object to a target
object via Transform operations.  With this, we can show that 2
Measurements are related via Transform, or an ImageAxis is a Transform on
the PixelAxes (or other ImageAxis).

Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dm/attachments/20180403/197839d3/attachment.html>


More information about the dm mailing list