spectroscopic data model

[CresitelloDittmar, Mark]

Gregory/Vandana,

There is a good bit of back story on the SpectralDM-2.0.  In the end
(2016), the project was scrapped for 2 main reasons:
  1) we were unable to get the necessary implementations to exercise the
model content.
      a) with a bit of sub-text that revision didn't really improve the
scope of the model (types of Spectral data supported)
  2) the multi-dimensional data model project became the priority

The idea being that the NDCube model would form a basis, from which
TimeSeries and Spectral models could become small specializations (ie:
'views' of a Cube).
The Spectral model made a step toward the current configuration, where
we've separated DatasetMetadata from the data (NDCube, etc).

The NDCube model has been pretty stable, but not well exercised.
The more recent focus has been on the underlying core models (
Measurements, Coordinates, and Transforms ) on which it depends.
   * Of these core models (Meas/Coords) have been through 1 round of RFC,
and are ready for another round, BUT we are again having trouble getting
application implementations to exercise them.
   * The transform model is nearly ready for RFC, we are more-or-less
waiting for the implementation team to complete their work.

We have prototyped a TimeSeries extension of the Cube model for TDIG to
play with.  There, we found that a very small extension of the Cube model
satisfied a wide range of time-domain cases.. so that is very encouraging.

As for a timetable, I expect my time to be returning to the Cube model and
updating it to the current state of the core models.  I suppose this is an
aspiration for the Autumn session.  Beyond that, the 'talk' has all focused
on the Time domain.. there has been very little discussion about the
Spectral model being a priority project.  I don't expect any movement in
that direction for the Autumn interop.  This is perhaps something which
should be discussed between the TCG and CSP.

Possibly more information than you really wanted..
Mark


On Tue, Jun 16, 2020 at 3:40 AM Dubois-Felsmann, Gregory P. <
gpdf at ipac.caltech.edu> wrote:

> Hi, Markus,
>
> I haven't been following the spectral model very closely until recently.
> Do you have a very rough sense of when a proposal based on the Cube and/or
> Dataset DM efforts could reach the content level of being ready to act as a
> successor to SpectralDM?  (Not including time to do final polishing,
> voting, etc.)
>
> Is this an aspiration for the Autumn (Northern) meeting?
>
> Thanks,
> Gregory
>
> ________________________________________
> From: dm-bounces at ivoa.net <dm-bounces at ivoa.net> on behalf of Markus
> Demleitner <msdemlei at ari.uni-heidelberg.de>
> Sent: Monday, June 15, 2020 11:44 PM
> To: dm at ivoa.net
> Subject: Re: spectroscopic data model
>
> Vandana,
>
> On Mon, Jun 15, 2020 at 04:30:07PM -0700, Vandana Desai wrote:
> > I have a question about IVOA spectroscopic data model.
> >
> > We at IRSA are starting a project that will require coding to a spectral
> > data model, and I was hoping for some clarification on the stability of
> the
> > latest SpectralDM.
> >
> > I see this:
> > http://www.ivoa.net/documents/latest/SpectrumDM.html
> >
> > which is V1.03 from 2007.
>
> This is an oversight.  It should be pointing to 1.1, and we'll fix
> this.
>
> > But this is listed as the most stable version:
> > http://www.ivoa.net/documents/SpectrumDM/20111120/index.html
>
> This is the current recommendation.
>
> > I also see this:
> > http://www.ivoa.net/documents/SpectralDM/
> >
> > which is V2.0 from 2016, but is listed as a "Proposed Recommendation".
>
> This was an effort to update the Spectrum DM to cover a few more
> cases, but it was eventually abandoned because (I'm simplifying a bit
> here) it wasn't quite in line with what the Cube DM would work out to
> be, and it was felt that efforts are better spent on the more
> general case which will eventually encompass what Spectrum DM 2 would
> have described.
>
> That the link beneath the document title in the document repo points
> to the abandoned V2.0 again is an editorial oversight, which we will
> fix.
>
> So, in short: Please implement against V1.1, and if you find
> shortcomings, by all means report them here. They will certainly
> inform work on the Cube and Dataset DMs.
>
> Thanks for reporting this,
>
>          Markus
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dm/attachments/20200623/4ac82b73/attachment.html>