# IVOA Northern Spring Interoperabilty Meeting 2017
### 15 May 2017 - DAL session 1

##### (notes by M. Molinaro)

## Keeping SCS up-to-date within DAL landscape
### (M.Molinaro)

Received some positive feedback to start up the revision of the
document. I will map the presentation topics into a wiki page or mail
thread to start feeling the pulse of the full community.

(DAL2 or closing remarks may give some more input)

## SSA revisited (Also SDM and SODA suggestions)
### (Petr Skoda)

Datalink plus SODA for spectra access does not solve all the issues. In
particular there are performance issues.

Query _POS_ parameter, also, is not sufficient, because there can be
sources withouth coordinates reference or name resolving may not cover
all of the possible objects.

Response format for spectra is set to VOTable or FITS/bintable,
otherwise one has to chooese native format. Also it has differences with
respect to SODA access.

Filenaming in the response is not covered by the spec.

SDM has some issue like wavelength conversion to/from rest frames and
also it would be useful to have a parameter to allow filtering on the
reference frame for wavelengths. High precision spectroscopy needs this
kind of things.

There are differences between SSA and SODA parameter value syntaxes:
this confused the users.

Theoretical use cases and things like continuum subtraction are not
taken into account.

## DataLink (Working on an implementation for stellar libraries)
### (Carlos Rodrigo Blanco)

What to do for use cases coming from the community that deal with
collections of spectra, catalogues, images and other ancillary files.

The presentation gave an example implementation of a collection that
includes catalogues, spectra, images and other ancillary files,
connected into a cascade system.

The idea is that if one can set up an SCS, an SSA and an SIA service, it
can also use DataLink solutions to put them together and help the user
to not loose track of the various sources s/he may be looking at.

## CASDA DAL Implementations
### (James Dempsey)

Update of the CASDA DAL landscape.

ASKAP data consists of images, spectra, cubes, momentums, with a 16TB
daily rate.

For spectra an SSA built on the SIAv2 experience but no DataLink has
been built. It includes DALI compliant interfaces.
Should SSA go towards DALI and SimpleDALRegExt compliance?

SODA is used to generate spectra out of cubes, with Montage to take care
of sub-cut and python to flatten the cube. The flatten system is not
done through a SODA triplet.

Future developments will involve: MOCs, HIPS, upload capability for TAP,
/examples endpoint and work on time series.

## Datalink and TAP in Aladin
### (Chaitra)

Presentation plus demo.

Client tested against CDS, CADC and GAVO Datalink services. There's an
issue with the way of automatically get the DataLink URL and give a
useful label to the menu entry in the Aladin client.

The Aladin TAP client generates ADQL queries, also gemìometry ones, out
of user input on the graphical interface. To do so it uses GLUE
dictionary records that describe the TAP services. It includes upload
capability.

## The efficiency of spatial indexing methods applied to large astronomical databases
### (Bruce Berriman)

Ongoing project (asking for contributions and hints).

First comparison of HTM versus HEALPix indexing in PostgreSQL using two
HW systems: a SunOS and a Windows. It used the 2MASS-PSC catalogue.

Steps involved creating the DBs and then querying them using various
index orders: 14, 8, 6, 4.

Scatter at lowest order produced by different densities at various
galactic latitudes (more dense data towards glat~0).

Performance differences most due to HW configuration than the index
order.

Next step to add a MS SQL server in the comparison and try higher than
14 orders.

Comments arisen on data clustering while preparing the data: it has an
high impact on the response serialization phase.