Minutes of joint DAL/DM session, 16-May-2006, Victoria

Chair DAL1/DM1: Doug Tody
      DAL2/DM2: Jonathan McDowell
Minutes: Markus Dolensky


slides are attached to 
http://www.ivoa.net/twiki/bin/view/IVOA/InterOpMay2006DAL


Agenda
1.1 Doug Tody - SSA Overview (25m)
1.2 Jonathan McDowell - Spectrum DM (45m)
-break-
2.1 Francois Bonnarel - Discussion on UTYPEs (20m)
2.2 Doug Tody - SSA Protocol (45m)
2.3 Markus Dolensky - Ranking algorithm (20m)
2.4 All - Discussion of SSA and Spectrum DM (25m)


Notes

Audience: 40-45 persons

1.1)
Q: Why does spectral data model accommodate SEDs but not echelle spectra?
A: Use cases treat echelle segments often as 1d spectra which are covered. 
   No consensus on how to associate spectra in case of echelle or in general.
Conclusion: 
- carry on with 1d spectrum for SSA V1.0
- sort out SED and echelle cases and general spectral data association in general thereafter

1.2)
Q: Quantum DM considered?
A: Not for this DM version.
Q: Is polarization vs. flux supported?
A: Not with this DM version. 

Parameterization of charac DM is needed for SSA interface.

Comment:
Against enforcing a certain serialization but to allow for a layer of data model.
A: to some extent possible through query result metadata
   
Options: continue with V0.96'ish DM - or - to go for STC/Charac. enhance V0.97
Comment: do not understand implications on interface - therefore in favor of V0.96
- some more discussion
Conclusion: beef up V0.96 limited to minor changes
 
- break

2.1
Q: Are possible ambiguities of re-used component models an issue
Q: Is a inference engine required?
Conclusion: 
  jCM: stick to current approach in spectral DM
  ... and hence no immediate change to VOTable required

2.2
Q: Version check necessary given the existance of a registry?

Q: getCapability method to return registry metadata
Some objection to add version number negotiation to service.
- no conclusion! needs to be hashed out with registry folks

Suggestion: 
At a parameter level one should specify whether a numeric range list 
should be monotonic and non-overlapping.

No objection on mandatory query parameters!




2.3 postponed to session DAL3

Jesus Salgado presents status of Simple Line Access Protocol
- comment: provide way to query for forbidden lines only
- comment: atmospheric lines vs. vacuum lines; one way to avoid conversion is to
  allow to specifying frequency instead of wavelength

Doug about SIA V1.1

Francois B. gives talk on various SIA issues V1.0 & 1.1 incl. extension mechanism

Francesco Pierfederici gives demo on WCS augmented graphics 
  and instrumental footprint service

Doug introduces a Java library that implements a DAL client
- this was done in context of the NVO summer school 2005


Valid target names need to be defined for TARGETNAME; task for semantics WG(?)

Q: Use lambda for SPECRES instead of m?
some debate

Q: Negative redshift?
A: Query interface to be less precise than returned data characterization.