OGSA F2F Session - 10 October 2007 - EMS
========================================

* Participants

AS = Andreas Savva (Fujitsu)
SN = Steven Newhouse (Microsoft)
HK = Hiro Kishimoto (Fujitsu)
DS = David Snelling (Fujitsu)
FB = Fred Maciel (Hatachi)
A? = Andrea  Merzky (Louisiana State University)
MD = Michel Drescher (Fujitsu)
DF = Donal Fellows (Manchester Uni)
SM = Stephen McGough (Imperial College London)
JR = Joel Replogle (OGF)

Phone:

* Summary of Actions

  ACTION: Replace BES with HPCP with intro paragraph to explain
          relationship between these.

  ACTION: DF to review RSS section of document

  ACTION: MD to provide necessary xml fragment for Source and Target
          for DMI

  ACTION: DF to re-arrange the figure to make it look more
          conventional


Minutes: SM

Lead: AS & SN

AS: Nothing direct to show

File transfer scenarios
----------------------

SN: Retro fit HPCP for BES

General yes

AS: Should we only change where it makes a difference? Leave alone if
no meaning difference.

SN: Will this confuse if we use both?

SN: Explain in intro paragraph

SM: State that HPCP is a subclass of BES

SN: Will this make people think you can only use BES for some bits?

ACTION: Replace BES with HPCP with intro paragraph to explain
        relationship between these.

SN: Purge out CDDLM?

AS: If we purge what do we put in its place?

SN: Need to remark that CDDLM is resting

SN: Leave senario but remove specification. Need to watch as we are
"endorsing" these specs.

AS: Make it a scenario which doesn't use CDDLM directly. Say how CDDLM
could be used without endorsing

AS: Two types of scenarios - those which are mapped to specs
                         - those where a problem is identified but no
                           solution is specified

SN: Conceptual and actual scenarios

RESOLUTION: Remove defunct specifications and describe all scenarios
as conceptual or implemetable

AS: ACS is another spec

SN: RSS - has it be reviewed by DF?

DF: willing to review

ACTION: DF to review RSS section of document

AS: Other issues in document: Activity credentials

SN: These are undefined. Place in JSDL, message body or header

AS: Make no statement on Security

SN: Yes

RESOLUTION: Implied model - though no statement about where these come from

AS: DMI EPR's - where specified - in JSDL

MD: Use RNS name which gets resolved

SN: Ravi - specific ftp, gsiftp
   Michel - in RNS
   Steven - in EPR

MD: JSDL consumer getting JSDL then taking this and mapping

SN: direct, RNS or EPR

AS: Which is the preferred way as this is what readers of doc will assume

MD: EPR and RNS

AS: If specific protocol then don't need DMI

SN: Isn't this a template? Population with instance info

MD: Resolution is to have EPRs in JSDL is good idea

SN: Works well and consumer will know what to do with it

RESOLUTION: Use EPRs in JDSL Source and Target for DMI transfer

SN: Do we need to profile DMI for use in JSDL?

MD: As long as using DMI name space then this is OK

ACTION: MD to provide necessary xml fragment for Source and Target for
        DMI

UML Sequence Diagram (RSS)
--------------------------

DF: Information System (IS) sends info back to different service than
that which calls it

DF: Separate service generated for making candidate set

AS: Is info from Information Service all in one go or in bits?

DF: This just shows communication not how data is communicated

HK: Feed all info from IS to EPS at start

DF: This could fit within the spec - info from IS gets fed into ISS -
no specification of how

AS: Restructure as putting these three together?

DF: Yes you can, in old versions EPS and ISS have been together, to
make return come earlier

ACTION: DF to re-arrange the figure to make it look more conventional