Difference: InterOpPlenary20031017 (9 vs. 10)

Revision 102003-10-22 - MarkusDolensky

 
META TOPICPARENT name="InterOpOct2003"
Minutes of IVOA Plenary Session, Oct. 17, 2003, 8:30 - 16:00

Chair: Robert Hanisch [RH]

Minutes: Markus Dolensky [MD]


Content:


1. Intro, RH (8.30)

  • General announcements
  • Mark Ackerman is studying IVOA as a sociologic phenomenon; feedback on process

2. Astro-RG, Nic Walton [NW] (8.35)

Reports about new Global Grid Forum (GGF) Research Group for Astronomy. First meeting took place at GGF9 in Chicago, Oct. 2003

3. Overview of IVOA, Peter Quinn [PQ] (8.40)

  • 13 members, sum of member project's budgets 25 mio. Euro in 3-5 years
  • There is a paper about VO Compliance: "Guidelines for Participation"
  • Roadmap: Defined in June 2002 in Garching
  • Key activities set at IVOA meeting in Seattle, Jan. 2003
Priorities
1 Registries
2 DAL
2 UCD
3 DM
3 VOQL
3 VOTable

4. Registry, Tony Linde [TL] (8.50)

  • Pushed documents about Identifiers and Resource Meta Data from working draft to recommendation level
  • Overview of Identifiers Document
  • Overview of Resource Metadata Document
  • XML Schema: current version 0.8.3; overview using XMLSpy toolkit
  • Interfaces Discussion Points: browsing, admin methods, query language, call mode, response form
  • Harvesting/Replication: For Jan 04 demos it was decided to support OAI httpGet with 'date since'

5. VOTable, Francoise Ochsenbein [FO] (9.10)

Proposed improvements to VOTable specification:
  • utype attribute
  • link to data model
  • contents of the attribute as "namespace:datamodel-parameter"
Q: Should we rename utype to mtype or something else?
A: Probably. It is simply a matter of getting used to it anyway.
  • Arrays of variable strings: allows several words in a table cell
  • FIELD element as a pointer
  • #no consensus yet
  • encoding attribute for TD element to insert binary data at the cell level, e.g. an image; no consensus yet
  • extensions to COOSYS element; move the actual definition outside the table; Note to developers: COOSYS may become deprecated; it could be replaced by a TBD 'utype' attribute Motivation: there may be different coordinate systems required in a single document; get rid of restriction to extrasolar coord. systems
  • character representation of numbers; discussion started
  • should FIELD and PARAM elements be merged
  • other tabular representations: BinX, CSV, TSV, ...
  • mimetypes
  • Suggestion: Add ucd and/or utype attribute also to TABLE element
  • Suggestion: Replace choices by substitution groups in VOTable schema
  • Suggestion: Use URIs for utype values since this gives namespace handling for free
  • Suggestion: Use a namespace so that votable elements can be used in other XML document formats

6. VOQL, Masatoshi Oishi [MO] (9.55)

  • Discussed SkyNode Interface Working Draft
  • 3 Agreements:
  1. layered structure for VOQL (VOQL1 web services; VOQL2 federation; VOQL3 SkyXQuery)
  2. there is a need to modify VOQL with respect to UCDs
  3. ADQL should NOT in general be used as a registry query language
  • Working document review should take place within a month
  • ADQL parser from JHU is available from Wiki

ADQL/_SkyNode, Wil O'Mullane"> 6.1 ADQL/SkyNode, Wil O'Mullane [WM]

  • after SkyNode V0.5 (current doc) it will be split into 2 docs: SkyNode 0.6/ADQL 0.6(?)
  • web services using new ADQL parser are already available
  • SkyNode Interface: It needs to return at least a VOTable
  • call for implementers
Issues:
  • Neighbors to be added to ADQL
  • Functions port
  • Asynch Query/Auth/MyDB (further in the future)
  • UCDs in ADQL divided: some queries could be UCD driven, others probably not A UCD driven query requires the registry to know how to resolve it.
    Note: To resolve a query by UCD may require another query against the registry. There is, however, no such registry query language support yet.
  • No conclusion whether UCD queries should be supported in VOQL1 or only in VOQL2+
Q: Is VOQL doing too much? How do data providers cope with it? Where does the data model come in?
A: layered approach

break (10.30)

7. Uniform Content Descriptors, Roy Williams [RW] (11.00)

7.1 UCD Session Summary, Clive Page [CP]

  • IAU thesaurus is not a perfect reference since it isn't maintained adequatly anymore
  • RW and Sebastien Derriere made suggestions how to go from UCD1 to UCD2s (refactoring, etc.)
  • Patricio Ortiz showed summary of UCD usage, issues and pitfalls
  • Tom McGlynn proposed a simplified UCD2 approach
  • Pierre Fernique showed how to use UCD1 to formulate queries in Aladin
  • There was broad consensus that UCD2 would make such queries easier and related efforts sould continue.
  • Waveband definitions should be aligned with similar defintions in Registry and Data Model

7.2 UCD (cont.), RW

  • Introduced Concept and Property idea for UCD2
  • UCD Working Draft V1.9.9 available
  • about 50% of the audience wrote already code using UCDs; mostly simple string comparisons
Let's find use cases:
  • What would be useful APIs to UCD; how to express template/action; etc?
  • Set of relevant queries: photometric, ???
Projects:
  • prototype the UCD2 structure; make matching function;
  • table metadata presentation: derive human-readable description from UCD set
Q: Who's responsible for documenting the meaning of a UCD?
  • Applications: Aladin, VOPlot, transformations
  • Several members of the audience advocate NOT to tackle coord. transformations in this context since this is prone to false assumptions. *How to proceed? UCD1 are de facto frozen until agreement on future path is found. The UCD steering committee has to become active on this issue and what to do about backward compatibility.

8. Data Model, Jonathan McDowell [JCM] (11.55)

  • Since last meeting in May there was a transition to true collaboration on data models
  • documents being worked on: use case document, observation data model white paper, spectrum data model
  • Goal: provide abstract description of data
  • provide DAL interface design
  • reports on quantity, interferometry, transforms
  • work on mini models for errors, quantities etc.
  • discussion of spectrum model with DAL
Q: Is the goal to describe each ucd/utype attribute in any given VOTable by some data model?
A: No.
  • It should also be possible to use other models of different origin to some extend if they are described in a certain format like XML schema.
  • Observation Model: comparison of 6 existing models
  • The comparsion revealed a lot of conceptual commonalities but differencies in terminology.

8.1 DM Use Cases, Alberto Micol [AM]

  • different points of view: query, visualization, instrument model, data analysis (coverage, WCS,...) , data packaging, data provenance, observing program/data flow
suggestion for priorities:
  • data packaging & formats
  • data coveratge
  • instrument model
call for feedback about use cases within the next few weeks;please send data model use cases to dm@ivoa.net.

8.2 Spectral Data Model, JCM

  • The model is for 1d spectra and SEDs, but designed to be extensible to multi-dim spectra.
Q: Data Models are rather low level: What about high level model(s)?
  • Suggestion: Define concepts early on ("Error", "Domain Mapping")
  • Suggestion: use notation "[TOPIC]" in email subject lines to make it easier to follow threads
  • Suggestion: use namespaces in UML diagrams which should ideally include author and DM (group name);

Lunch (12.45)

9. Data Access Layer, Doug Tody [DT] (14.00)

  • introducing virtual data
  • Dataset IDs: need to define those
  • utype, UCD normalization needed to reference data models
  • Proposal: access reference templating to shrink large SIA/SSA result tables
  • formal getData method; could use XPath
  • Major topic: Simple Spectral Access (SSA)
  • A survey among data providers and consumers has been conducted.
  • develop uniform data model with data modelers for 1dspectra and SEDs
  • provide uniform interface to 1dspectra and SEDs
  • service interface similar to SIA, Cone Search
  • output formats: text, VOTAble, FITS, graphics
Q: What formats are required to support?
A: Not yet decided.
  • Suggestion: A service only needs to provide one format. Client first looks up registry to find appropriate service that returns spectra in desired format.
  • SSA interface issues
    • Registry integration: service metadata query, SSA service metadata
    • SSA service verification - should be part of registration process
  • Query: discussed set of query attributes; simple keyword queries (now), ADQL queries later
  • initially supported query params: pos, size, spectral resolution, bandpass, time
  • Suggestion: Normalize services soon (SIA,CS,SSA)
  • fits (table or image?); 1d image containing linear spectra would avoid duplicate effort when defining also a VOTable format
  • Suggestion: Do not invest much in a FITS format because FITS standardization for spectra is under way.

10. 14:50 Grid & Web Services, Guy Rixon [GR] (14.50)

  • metadata interface between registry and service:
  • Suggestion: rename metadata port to resource
  • Wil O'Mullane to write up and liase with registry
  • Heartbeat interface: availability of service
  • TCP/UDP ports: firewall issue
  • GR to maintain list of ports (GridFTP, LDAP, ...)
  • services to standardize: name resolver, "Network weather", unit conversion,
coordinate conversion, authorization/access policy, storage for derived data-sets.
  • list of services: SkyNode VOQL service, MySpace data storage, GGF authorization port-type(s), ...
  • access control: authorization, authentication, ID certificates
  • NW to get legal advice about storing personal information
  • There's a variety of Grid service usage scenarios with different levels of possible support
  • no consensus on tools to use
  • Suggestion: Set up wiki page where we can share experience with various environments/tools (Axis, C#, ...)
Q: How to connect web services to old legacy software like Fortran tools?
A: See Martin Hill's posting to DAL mailing list.

11. Wrapping Up, RH (15.30)

  • post implementation plans, schedules of WGs to wiki
  • next IVOA telecon with WG leads
  • cross talk between groups can be improved
  • splinter sessions are big but still such that everbody can express opinion
  • Mark Ackermann suggests to define items that are on the IVOA critical path

Meeting closed at 16.00

Calendar

  • January 2004 demos at AAS for NVO and at ESO (most likely venue at present) for AVO
  • Atlanta Meeting after AAS
  • next interop workshop: Boston, Apr/May 2004
  • 2nd next interop workshop: Pasadena, Oct 2004
  • bookmark this page for updates

Action Items

Changed:
<
<
  • Action IM1003-P2-1 FO: priortize VOTable changes to work on; consider needs for next Jan 04 demo prepare working draft V1.1 by mid-Nov. for simple important changes transform V1.0 doc into IVOA proposed recommendation by mid-Nov.
>
>
  • Action IM1003-P2-1A FO: priortize VOTable changes to work on; consider needs for next Jan 04 demo
  • Action IM1003-P2-1B FO: prepare working draft V1.1 by mid-Nov. for simple important changes
  • Action IM1003-P2-1C FO: transform V1.0 doc into IVOA proposed recommendation by mid-Nov.
 
  • Action IM1003-P2-2 UCD technical board: Decide on how to make transition from UCD1 to UCD2 and how to deal with backward compatibility
  • Action IM1003-P2-3 WG leads: post schedules to WG wiki pages
  • Action IM1003-P2-4 UCD/Registry/DM: align waveband definition

<--  
-->
META TOPICMOVED by="MarcoLeoni" date="1066586402" from="IVOA.InterOpPlanery20031017" to="IVOA.InterOpPlenary20031017"
 
This site is powered by the TWiki collaboration platform Powered by Perl This site is powered by the TWiki collaboration platformCopyright © 2008-2024 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback