[dais-wg] Notes from the 06/12/05 DAIS telcon.

[Mario Antonioletti]

Hi,
    These are notes from the last DAIS call. Note that there will be 
another DAIS call on Tuesday. Details at:

http://www-unix.gridforum.org/mail_archive/dais-wg/2005/12/msg00003.html

 			Mario

+---

DAIS Telcon - 06/12/05
======================

Chair: Norman Paton
Notes: Mario Antonioletti

Attendees:
 	Dave Pearson, Oracle
 	Simon Laws, IBM
 	Mario Antonioletti, EPCC
 	Norman Paton, University of Manchester
 	Amrey Krause, EPCC
 	Susan Malaika, IBM

Agenda:

  1. Review of Open Specification Issues on GridForge.
  2. Feedback on latest versions of specifications.
  3. Forward planning for DAIS-WG.
  4. DAIS session planning for GGF16.
  5. AOB.

+--

SL has gone through the core, relational and XML to make them
consistent. The WSDL is in-line with the documents.

There are differences in the WSDL that Dave Vyvyan has picked up while
implementing the WS-DAIR spec. SL will have a look.

NP has done a pass through and made some typo changes but overall
the specs are looking quite good.

1. Review of Open Specification Issues on GridForge.

Look through the open issues in Grid Forge. Start with the WS-DAI issues.

Summary #972: Exposing access privileges.

NP: there is also the question of how much of the CIM model we expose.

SM: CIM discussion is still in progress. They had not resolved the
     namespace issue yet. Will try to get the current status.

NP: overall access to the CIM model is outside our scope.

SM: WS-CIM in DMTF define the representation.

MA: do we have to say anything about accessing the CIM model then?

NP: we may need to revisit this - need to check what they provide.
     Close this issue for now by stating that this is out of scope for
     this specification.

status: closed.

Overall access to different aspects of the CIM Model within web
services is outside the scope of DAIS.  It is believed that another
group within the DMTF will define web service access to CIM as a
whole.

  	977: Disconnected Data Sets.

NP: change this to pending, and state that the functionality will be defined
     in future versions of the DAIS specification.

status: pending.

The provision of enhanced functionality around disconnected data sets,
such as using them for updates or for writing back, is considered
beyond the scope of the first version of the DAIS specifications.


  	1101: Review Faults proposition

SL: our view of faults has changed ... no faults are mandated by the
     specs ... they are recommended and it is extensible. We do not do
     anything outside WSDL. When WSDL 2.0 comes out we can review this.

status: closed.

In the core spec we now define a recommended set of faults in three
categories:

   General
   Direct Data Access
   Indirect Data Access

These faults are included in DAIS compliant interfaces, when required,
using standard WSDL fault mechanisms.

      	1102: What security hooks are required

NP: at the moment we do not define any. We see this as being out of scope
     and specifications will be used from elsewhere.

status: closed.

The expectation is that the implementers of DAIS services will use
specifications from elsewhere for conveying or accessing security
information.

         1196: Scope of DataFactory in Specification

SL: have factory operations but we do not factories for factories.

NP: Out of scope for the present version of the specs. Make it pending.

status: pending.

This is considered relevant but out of scope for the first version of
the DAIS specifications.

         1343: Provide a default data resource for a data service.

SL: There are no default data resources.

status: closed.

We provide a service level message that allows the list of valid data
resources to be retrieved. If no data resource id is specified when a
message expected then a fault is returned.

  	1344: Need to ensure that WSDM's MUWS and MOWS could be used for
               data management.

NP: will convert this to pending. Can be reviewed for some future version
     of the DAIS specifications.

status: pending.

Use of such other specifications for data resource management is out
of scope for first DAIS specifications.

  	1370: Provision to write back through DAIS services.

SL: there is nothing in DAIS that pushes the data to a service.
     What we do not allow is for the data to derived from another service.
     There is an issue that needs to be more carefully worded.

status: closed.

DAIS specs don't forbid such updates.  Pending issue 1707 relates to
the passing of references to update messages.

Moving on to relational issues:

         1197: RelationalSchema Informational Property

NP: should review this when the CIM model position stabilises.
     Leave as open.

status: open

Review when CIM database model stabilises.

      	1198: SQL RowSetSchema

NP: this is separate as the first one was the metadata property and this
     was the result type ...

SL: we have flexibility for result types ...

NP: so we can close this one ...

status: closed.

Resolved - specification is flexible in terms of return types.

No XML issues.

2. Feedback on latest versions of specifications.

NP has some issues on the specs. WS-DAI, p8:

o A message for translating from a data resource abstract name to a
   data resource address. This address MAY be the EPR of a WSRF data
   resource if the WSRF data resource interfaces are in use.

NP: this should be a SHOULD or a MUST.

SL: needs to be a MUST.

NP: last sentence of Section 5.0:

Where a WSRF data resource is present the messages associated with the
WSRF data resource MAY also be used.

NP: should this be a MUST?

SL: should probably be a MUST ...

NP: In section 7.1.3:

The DataResourceAbstractName passed in each DAIS message simply
duplicates this information and may be ignored by an implementation.

SL: advantages of checking is that you show some consistency ... I would
     keep it like that but make it a MAY...

...

NP: XML 5.5.1:

o ColllectionName: An OPTIONAL parameter that contains the URI of the
   collection from which the subcollection will be removed. If no
   collection name is provided the top level collection is assumed

typo in CollectionName, also " from which the subcollection will be removed"
should probably not be there ... this is producing a copy ... or does it have
....

question is what it should do.

SL: sensitive corresponds to the parent and not how the child affects the
     parent ... if we don't say it's a copy then it is logically inconsistent
     but we have not said it's a copy. Cautious about saying it's a copy...

NP: let's not resolve this today; we can think about it offline and
     review at the next call.

SL: maybe worth putting a tracker about upwards propagation...

NP: on p15

     "o	An XQuery expression - The request must conform to the XQueryX schema."

    Can this just be a string?

AK: yes.

NP: so, I will change this accordingly and make this explicit ....

SL: have some things that have not been addressed that we wish to put
     as a future.

...

NP: Scan through the remaining issues for next week ... do forward planning
     next week. Need to do GGF forward planning now though.

     Need to provide time for:

     Will have submitted the specs, gone through the editorial thing
     and entered the public comment by GGF. Should allow some
     opportunity to discuss open issues go through any public
     comments. Also, describe the work on the implementations. Steve
     will Lynden will be travelling to GGF...

...

DP: should we have one or two sessions?

...

NP: I think we'll need 90 minutes ...

     Update to the comm 5min
     Feedback, 30 minutes
     Implementations, 30 minutes
     Futures, 20 minutes

     Do some more issues next week.

+-----------------------------------------------------------------------+
|Mario Antonioletti:EPCC,JCMB,The King's Buildings,Edinburgh EH9 3JZ.   |
|Tel:0131 650 5141|mario at epcc.ed.ac.uk|http://www.epcc.ed.ac.uk/~mario/ |
+-----------------------------------------------------------------------+