From: Carleton DeTar <detar@physics.utah.edu>
To: cmendes@cs.uiuc.edu, djholm@fnal.gov, edwards@jlab.org, watson@jlab.org,
        rdm@phys.columbia.edu, brower@ctpup.mit.edu, avp@lns.mit.edu,
        simone@fnal.gov, sg@fuji.physics.indiana.edu, osborn@physics.utah.edu,
        doug@klingon.physics.arizona.edu, gregory@physics.arizona.edu,
        mdp@fnal.gov, petrov@phys.columbia.edu, chulwoo@phys.columbia.edu
CC: detar@physics.utah.edu
Subject: Minutes August 29, 2002

	   Software Coordinating Committee Conference Call
		       August 29, 2002 3:00 PM EDT
			  Recorder: C. DeTar

Present:

  Edwards, Holmgren, Simone, DeTar, Osborn, Mendez

Absent:

  Mawhinney, 
  Kostya Petrov, Chulwoo Jun, Brower (CERN), 
  Pochinsky, Watson (Scotland ?)

** Action items

- ----------------------------------------------------------------------
Agenda:

Initial items:
1) Update on QDP C Level 2 API (DeTar)
2) Update on QDP C++ API  (Edwards)

The main item on the agenda is I/O. Subjects include:

1) Programming interface - a QDP (level 2) data-parallel model
   has been suggested
2) File formats
3) Web access/cataloging/archiving

NOTE: These 3 items could/should be logically separated. However,
we've not had any serious discussion lately as a committee
on any of the points, so some discussion and possible interplay
is warranted.

This call should hopefully provide a start for the work
of the various relevant subcommittees devoted to I/O & Data as
indicated by Richard's Task Chart.

----------------------------------------------------------------------
Edwards: I/O issues
    Data parallel interface
    Header information
    File formats
    Web cataloging (XML)

Simone:
    Want minimal set. Size of atomic type, types, serial order
    Rest is XML.  Extensible, but not required.
    NERSC is overspecified.

DeTar:
    Should we encode the XML at the head of the file or
    in a separate file?
    What about a separate XML file for the entire configuration set?
    Then we don't need to repeat it in every file.

Edwards:
    Readers for binary data are not very sophisticated.
    Should we be using XML at all?
    That is what the Particle Data Grid project is working on.
    Chip is visiting Edinburgh working on scheme for lattice data.
    Wanted web services.

DeTar:
    Can we get a document describing their current plans?
    
Edwards:
    A common web reader is SOAP.  

DeTar:
    Do we need to make all the XML information to the program
    reading the data?

Edwards:
    Perhaps we need it to at least for a consistency check

DeTar:
    So if we don't want to set standards for what items are in
    the header, the user has to be flexible about inconsistent
    choices of items.

Simone:
    If we keep it simple, say keyword, value pair, then
    a simple parser can handle it.  

Edwards:
    Fortran namelist scheme allows some nesting.

DeTar:
    Formats:
    (1) lexicographic single file
    (2) dump to multiple files
    (3) checkpoint single file - arbitrary order, but with an index
    Do we want this.

DeTar:

    For QDP need to handle the special case of writing four gauge
    field objects at once (for the four coordinate directions).  Then
    we should be able to write single gauge field objects,
    particularly propagators and scalars.  Some of us also use arrays
    of spinors (e.g. eigenvectors).

Result of discussion among DeTar, Edwards, Simone:

    As for the XML file, we need a user call in the interface for
    writing keyword, value pairs to the file and a prescription for
    when this is supposed to be done.

Edwards:
    I/O Subcommittee assignment:  
       Data file formats and QDP interface -- DeTar, Edwards, Simone
**       Jim to put together a paragraph proposal


    
----------------------------------------------------------------------
Conference concluded at 4:50 PM EDT.
Next plenary call Sep 5 3 PM EDT.