SDM Running Meeting 14/06/2022

Meeting Outcomes

• In order to be able to map NED SEDs or Spitzer spectra on the spectrum data model, it is necessary to add the support of lower/upper limits as well as spectral orders.
• The current SDM has been published in 2011 with document standards of that time.
  • We have no resource to upgrade it totally
• Some others requirements have been proposed but we have decided not the tackle with them since they have no impact on the current operations.
• It has been decided to move toward SDM1.2
  • Minimise model changes
  • Not to VODml-ze it
• Upper/Lower limits:
  • add 2 attributes to the Accuray class
• spectral order
  • Add 2 attributes the SpectralAxis class
  • Absolute order / relative order
  • The purpose of these 2 orders must be clarified
• As the changes are small, we can expect a smooth RFC
  • We need at least 2 reference implementations
  • We need one validator
  • We need the document to be updated
    • use-case
    • field description
    • diagrams
  • A list of actions has been setup (see below)

Actions

#1 LM: working group poll

Ask the working group whether our roadmap is workable

#2 FB: contact Cassis team

Ask JM Glotrian whether he volunteers to implement SSDN1.2 on Cassis to provide one reference implementation

#2 LM: Contact Markus

Ask Markus whether he volunteers to contribute to the reference implementations:

• as data provide with some DACHS instance
• as client with SPLAT

#3 VD provide spectrum/SED sample

Provide data sample (either spectra or SED) for which the new attributes are relevant.
This sample should be accessible through an online service.

#4 LM/MCD: merging SDM related Wiki pages

Making more accessible the different pages related to the project

#5 VD: provide use cases

Provide use-cases that justify the move to SDM1.2

#6 MCD: edition

Document edition:

• Add descriptions of the new field.
• Replace blue diagrams with the UML diagrams generated by MCD (it is to be noted that original blue diagrams are not perfectly compliant with the model XSD).

Etherpad Notes

Participants (7): Mark CD, Laurent Michel, Jesus Salgado, Francois Bonnarel, Vandana, Marco Molinaro, Tom Donaldson

*Documents*:

* Vadana's talk (May 2021) pdf
* MCD's overview (Fall 2021) pdf
* SDM specification: pdf
* SDM schema: ivoadoc/xml
* SSAP specification pdf

**Model enhancements suggested by Vandana (IPAC) in 2020 *

**Topic 1*: Spectral order support

* 2 orders: absolute and relative
* VD proposal
* one table per order
* add a column
* MCD proposals (to add a column):
* SpectralAxis attributes
* Specific class to be linked with either SpectraAxis or DataAxis
* Q: Do we need to have orders on other axis than SpecralAxis?

*Topic 2: *Upper/lower limits

Error max taken as Upper limits (doable but not clean)
* No lower limit support
* MCD
* Add fields to Accuracy class
* Consider as a specific measure type

*Topic 3*: VOUnit compliance

* Moving toward VOUnits (instead of OGIP see page 15)
* What is the impact on SSAP (more restrictive than SDM)

* in cases where details of the data model differ between SSA and the SDM, the SSA specification takes precedence. (SSA P11)*

* Topic 3+:See Vandana slides (A bit far from SDM)....

**Context and Issues (in bulk)**

* Old standard, technically difficult to upgrade
* How not walking on CubeDM toes
* Should we move toward SDM 1.1 or just make an endorsed note?
* Where is SDM used?
* What about SSA responses
* Usually refer to FITS files
* Is there with VOTable references?
* Should we VODMLize it?
* Writing/Implementation/Validator
* Who
* When
* model extension (6.3 p 43); *For the FITS serialization, implementors may add arbitrary additional keywords or table columns; readers must be able to handle files containing extra keywords and columns, and are encouraged to propagate such extra information when copying files. This permits local conventions to be layered on the basic definition.*

**Actions:*

* checking compliance OGIP/VOUnit
* FB contact CASSIS (ref impl)
* LM contact MD (ref impl)
* LM confirm that people do prefer SDM1.2
* LM/MCD merging Wiki pages
* MCD replace blue diagrams with UML ones

**Meeting minutes:**

LM: Process IVOA to accept proposal and find possible approach to include them

- Support spectral order
- upper/lower limits
- Other

Focus on the first two points as they are doable

MCD presented proposal to answer these problems

- Utype modification is adopt for the spectral order. Proposal to add order and relorder attributes to SpectralAxis

(better than adding an order class)

FB: order and relorder are not redundant

Vadana: it could look redundant but spectral data make the distinction

- For upper/lower limits, add UpperLimit/LowerLimit to accuracy (proposal 1)

Second proposal is to add inside sedLimit but this option looks overcomplicated

FB: About units, most of the data is in fits format?

Vandana: Many formats, including ASCII

JS we could state the units in new SDM must by VOUnit compliant

MCD: What can be the reference implementations? One from Vandana but the other?

Vanada connected with a group implementing a python library to parse spectral data. Possible second implementation on python?

LM: Contact the client developers to check if they could modify the libraries to provide support to these new features

Also server providing these new data

MCD: Gavo?

MCD proposes just an endorse note "IPAC extension to Spectrum-1.1" (more than a more official Spectrum model RFC process)

Start to create content on the wiki pages to follow-up the effort (some pages created but still empty)

MCD: Diagrams are not really in line with the schema. What to do?

Probably replace diagrams in line with current ones

LM: Moving to VO/DML (as seen in PhotDM) is not easy. To skip this point in this case would be nice