CommentsOnAdassReport

 

Thanks Anil Chandra for reporting so effectively on the ADASS meeting 2004. I have a few comments:

  • In the introduction you state that in Australia we have no data archives accessible through a VO-compliant interface. Actually, HICAT (the HIPASS catalogue), the HVC catalogue, the HIPASS Bright Galaxy Catalogue, and the SUMSS Catalogue can all be queried with a ConeSearch. While technically not an endorsed standard of the IVOA, it truly is a de facto standard, and delivers VOTables to users in a defined manner. n.b. we expect to have a SIAP service up and running in Melbourne within 30 days.
    • I stand corrected. I was well aware of these ConeSearch interfaces so I'm not sure why I overlooked them. - Anil Chandra
  • Indeed we really do need to set up a registry in Australia. Unfortunately there is no decent "reference implementation" of the registry standard and that leaves us either having to code our own (wasteful) or wait for someone else to (somewhat frustrating). The need is definitely identified, but the path forward is not.
    • There are a handful of registries that implement the OpenArchivesInitiative interface, allowing the NVO registry (http://nvo.stsci.edu/voregistry/index.aspx) to harvest their records. If there is a significant overlap between the OAI standards and the IVOA's registry standards, then maybe we should look at starting with OAI. -Anil Chandra
  • You comment that OpenSkyQuery and ADQL are possibly too immature for us to consider at this stage. I think this is right, but I think we need to recognise this area as one of the most important in the VO effort and we urgently need someone in the Aus-VO community to be tracking if not contributing to these efforts. Unfortunately I was unable to spend a planned month in Japan this year to do this. If the developments of OpenSkyQuery and ADQL lead to dependencies on databases (eg. via SQL) we need to know about it and be developing the appropriate expertise here.
    • I agree that we need to track the developments in this area. I would be very surprised if the OpenSkyQuery and ADQL standards themselves have dependencies on databases. However, having a database backend seems like the natural way of implementing this kind of thing. I will not be surprised if mappings between ADQL and SQL begin to emerge as these standards mature. -Anil Chandra
    • The NVO SS package found at http://chart.stsci.edu/twiki/bin/view/Main/Software contains code to setup a basic SkyNode. It appears reasonably straight forward. They provide Java code which you need to interface with a database such as MySQL. This is something that I could conceivably work on, it is somewhat related to building a machine learning matching tool. Providing HICAT on this service for example is certainly a modest goal. The main thing that I have noticed about this service is that it is priciply dealing with the distributed nature of the join (unlike CSIRO ICT who focused on the spatial nature of the join) The size of output is limited to 10000 records at the moment, which in my view is a very serious limitation-- David Rohde
  • VOSpec is listed as a "very nice tool". My experience of it (early 2004) was dreadful. It appeared to function "ok" for optical spectra, but simply refused to handle spectra from HIPASS because of the different spectral axis (frequencies of order GHz c.f. wavelengths of order nm). At the time it offered some basic Gaussian fitting capability, but that's of virtually no use in many domains.
    • Interesting! I must admit, I formed my opinion based on the demonstrations that were given. Nevertheless, even if the tool itself is not scientifically useful yet I see it as a good demonstrator for the usefulness of the SSAP. If you haven't done so already, I would encourage you to raise these issues with Pedro Osuna, who I believe is looking after the product. -Anil Chandra

-- David Barnes - 02 Nov 2004