Minutes of Teleconference 10 October 2006

Data Access and Prototype WG (IPDA)

 

Date: 10 October 2006, 15:00-15:50 UTC

 

Participants:

 

            PDS-EN, JPL:                      Dan Crichton, Sean Kelly

            PDS-Geoscience Node:  Susan Slavney, Ed Guinness, Tom Stein

            PSA-ESTEC:                                   Joe Zender, David Heather

            PSA-ESAC:                                     Christophe Arviset, Jesus Salgado (c)

 

0     Action Items

 

Actions related to Data Access WG taken from IPDA general actions.

 

AI-NO

Actionee

Description

Status

200605-02

SK, JS

Update existing agency prototypes to support for spatial response (15 June 2006)

2006Jun22: done on PSA side (closed), SK will do before end of June 2006

PSA: closed

PDS: ongoing

Superseded by new action

200610-01

200605-03

ALL

Develop individual clients to demonstrate interoperability:

. Map-based demonstration (for mid July) as IDL search and map-based search

. Integration into the PDS-D search (30 June)

Map-based done until new PDS data available

PDS prototype ongoing accessing PDS datasets through OODT and PSA datasets through PDAP. Not public yet

Superseded by new action

200610-01

 

 

1     PDS and PSA PDAP compatibility versus map client implementation (SK+JZ+JS)

A nice to have during the next November meeting is a PDS-PSA interoperability concept demo using the PSA-ESTEC map client.

 

Testing PSA-PDAP implementation:

- Some PSA Server access problems due to different reasons: connectivity problems (hopefully solved) and performance degradation (problem identified and to be solved for next version)

-         VOTable dtd pointing to version 1.0 instead of 1.1 (to be changed). The parsing is disabled in the client so this is not a showstopper, but it will be changed for consistency

 

Testing PDS-PDAP implementation:

-         Empty VOTable from PDS server for certain queries. SK checked the query directly and he found to possible problems: geometrical information could be wrong (all products look to be in the same spatial position), and memory problems (too many objects). This second problem could be solved using pagination in a future PDAP document version including pagination, but it will be solved in a near future in two possible ways: limit the number of products to a certain number (only to be used for the November meeting) or stream the VOTable in bunches instead of loading the entire result in memory. Both solutions are also applicable for the PSA-PDAP implementation

-         JZ reported no footprints from the PDS-PDAP implementation have been displayed yet. JZ asked if the RESOURCE_CLASS=IMAGE queries have been implemented and SK answered affirmatively

 

As some items needed to be changed in both servers and, perhaps, in the client a new action is raised on SK, JZ and JS to solve/trace all the open issues for the demo.

           

2     PDS PDAP client status (SK+DC)

SK reported a client has been implemented connecting to PSA datasets through PDAP protocol and to PDS through OODT. DC confirmed there are plans to present this client during next November meeting as a different flavor of client.

 

3     PDAP document suggested changes discussion (JS+SK)

Some comments/changes proposed to the PDAP document (by SH and SK). This is the summary and status:

-         POST requests as well (Status: accepted)

-         URL spaces allowed but replaced with the + character (Status: accepted)

-         Reference to VOTable version 1.1 instead of just VOTable (Status: accepted)

-         UnicodeChar instead of char: Accept characters outside US-ASCII range for dataset IDs. (Status: This could produce parsing problems for clients. For the first protocol version, we can maintain this constraint and, in case other agencies need special characters, it can be modified in a future version)

-         Typo POLIGON instead of  POLYGON in example (Status: accepted)

-         RESOURCE_CLASS instead of RESSCLASS as the name for the field for the output (Status: accepted)

-         DATA_SET.DATA_SET_NAME instead of DATA_SET.DESCRIPTION (Status: accepted)

-         PRODUCT.PRODUCT_ID instead of PRODUCT.DESCRIPTION  (Status: accepted)

-         RETURN_TYPE=TEXT instead of RETURN_TYPE=ASCII: The main objection is not because of the parameter name but because of the limitation to US-ASCII characters implicit in the name. (Status: As it is better to have this limitation for the parsing in clients, this can be maintained for the time being. In case other agencies need special characters, it can be modified in a future version)

-         Use of RETURN parameter instead of RESOURCE_CLASS=METADATA. SH proposal to include a new parameter (RETURN) to be used for discovery, leaving RESOURCE_CLASS for its normal use. (Status: accepted. Action to SH to write down the proposal in a way to include it directly in the PDAP spec)

-         Use of RETURN to access metadata field in a fine grain: SH proposal to use the RETURN parameter to access specific metadata fields (e.g. RETURN=DATA to access the data access link) (Status: Following this approach, the PDAP will have a fine grain access to metadata, so it can be very useful. However, it has to be evaluated if that implies complexity in the PDAP server implementation (filtering could be done at server, at client side or both). Open for discussion in the forum)

JS will prepare a version containing all the accepted changes and it will be sent again to the DAP WG forum

 

5     AOB

 

6     Summary of New Actions

200610-01 SK/JZ/JS Coordinate/implement necessary changes in PDAP server/client map implementation to be presented during November meeting

 

200610-02 JS Add accepted changes in the PDAP spec doc

 

200610-03 SH Write down use of RETURN parameter to query service capabilities (QUERY_PARAMETERS) to be included in the document

 

6. Next teleconf

Next teleconference is planned for 31 October 2006, 16:00 UTC,  (17:00 ESTEC/ESAC, 08:00 JPL/PDT) just after the Archive Standards Working Group teleconf, e.g., at ~16:40