OGSA Teleconference - 19 October 2006
=====================================

* Participants

  Abdeslem Djaoui (RL)
  Michel Drescher (Fujitsu)
  Chris Kantarjiev (Oracle)
  Hiro Kishimoto (Fujitsu)
  Andreas Savva (Fujitsu)
  Dave Snelling (Fujitsu)
  Elen Stokes (IBM) 
  Jem Treadwell (HP)
  Jay Unger (IBM) 

  Apologies: Takuya Mori

  Minutes: Andreas Savva

* October 16 minutes approved with no changes

* Action item review

  [Only items with updates are listed below.]

   - AI-1009d: Topic list posted. Closed. (Note that there were two
               items with the same number on the agenda, both closed.)
   - AI-1005a: Postponed discussion until Alan Sill is on the
               call. (There was some confusion since again there was a
               second AI with the same number.)
   - AI-0921c: New ADs have been assigned. Dave S has not approached
               them yet.

* Nov. F2F update

  There is updated information on the wiki.

  It should be possible to book the Doubletree at the IBM rate via the
  Hotel's reservation center.

  Agenda discussion focused on the possibility of working on
  Information Model/XQuery/JSDL 2.0 topics, especially looking at the
  expected participants. (It may also be possible to get some IBM
  participants from Italy.)

  Andreas had already asked on an earlier JSDL call about arranging a
  session at the OGSA F2F, and the response was positive. An agenda
  has not been discussed yet.

* Information/data modeling (XQuery discussion)

  Jay re-stated the original intention of using XQuery expressions as
  requirements. He acknowledged that since XQuery is a programming
  language the possibility of 'doing harm' exists. For example, using
  unbounded recursion (maliciously or accidentally); calling outside
  functions; or through 'excursions in URI space'. Such concerns are
  however profiled(?) as implementation concerns in the base XQuery
  specification. It should be sufficient to apply common techniques
  such as time-bounding queries to address these. Some guidance could
  be provided in the spec but should stop short of requiring specific
  mechanisms. (For example, there should not be any specific
  requirement to use document collections in queries; it should be
  implicit.)

  Michel has done some further work, also consulting with people
  actively involved in this area. He has modified his position since
  last time and now thinks that the original approach is fine. He has
  some working examples/code with dynamic xqueries and so on.

  The consensus is to continue on the current path with XQuery as a
  requirements language.

  There was also agreement that for interoperability/ease of adoption
  defining a profile on XQuery with those features that would be
  useful to job descriptions would be a good idea. A base profile
  could restrict the features that may be considered more dangerous or
  less necessary. Since there are already fairly complete XQuery
  implementations this may be more guidance for clients writing
  XQueries rather than servers processing them.

  The plan is to continue this work on future calls and at the F2F.

* OGSA AuthN WG BoF draft --- Postponed no quorum
* Security Profile Update --- Postponed no quorum