Apologies for the lateness in posting this set of minutes out.
DMI Telcon - 17 July 2009
=========================
Present
-------
Mario Antonioletti, EPCC
David Meredith, STFC
Shahbaz Memon, FZJ
Peter Turner, University of Sydney
Gerson Galang, VeRSI
Agenda
------
- Agenda bashing
- Progress update
- MAA determine if proposed changes can be passed off as errata.
- AL determine if existing spec can transfer multiple directories
living on the same host.
- SM WSRF OGSA-DMI rendering.
- Other stuff ...
- Planning
----
Actions:
[MAA] To come up with a template document to explore if required
functionality can be put in a spec addendum/new spec.
[MAA] Point folks to the Plain WS Rendering of the DMI spec.
The document that went to public comment (and went down a black
hole) is available from:
http://www.ggf.org/Public_Comment_Docs/Documents/2009-01/ogsa-dmi-wsi-rende…
The source for the document is at:
http://forge.gridforum.org/sf/go/doc15259?nav=1
[MAA] Look at Shahbaz's WSRF DMI rendering.
DONM: 8am UK time (GMT+1) Friday 11 September
Agenda bashing:
--------------
Take as is.
Progress Update:
----------------
o Mario on changes to the spec:
Basically errata works within the first 6 months the document is
out on the filed. Current spec has been out for about a year now
so we are at the discretion of the OGF Editor.
May be ok to add the extensibility points as is - don't think
anyone has really implemented the specs afaik. Other changes
may require a new process.
Might be good to try to do an addendum to the DMI spec that
has the requested changes.
http://forge.gridforum.org/sf/go/doc14287?nav=1
o Allen is in the US currently so this time is not suitable for a
telcon:
Allen not had time to look at whether multiple data transfers
from the same host supported by current version of DMI. Allen
somewhat uneasy about bloating the DMI spec.
o Discussion
Mario thought it might be good to see if the extensibility points
required by DataMinx could be retrofitted into the existing spec
and the new required functionality put into a new spec addendum
or at least see how far we could get with that. There are too
many possible details in the proposal that David put in:
http://www.ogf.org/pipermail/ogsa-dmi-wg/2009-July/000493.html
We do not have to solve the general case, but enough to accommodate
the DataMinx requirements.
Mario asked if there would be interest in doing this. Shabhaz
said that UNICORE could do one of the implementations and was
wondering whether DataMinx could do the other.
There was a cautious maybe from the DataMinx folks.
It is important to have an implementation in tandem with the
writing of the spec in order to ensure that there is some
sanity. It would be good if this effort was driven by end
user requirements.
Peter is uneasy about having to steer a path between JSDL and
DMI for the DataMinx project. JSDL takes a computational view
of things - can cheat and have a data movement job by using
the data staging in and out and a trivial job such as:
stage data in
/usr/bin/date
stage data out
which would effectively do the data movement in a roundabout
way.
Would be good to try and get some kind of inter operability
between JSDL/DMI so the two could operate properly (if possible).
Gerson stated that the DataMinx Schema was using bits from
DMI and JSDL and that the DMI may care to look at this and
comment.
There was a discussion on the functional spec (which just
contains the top level schema) and the renderings that
contain the actual WSDL. The plain WS rendering went to
public comment and then disappeared - Mario chasing up
with Greg Newby.
Shahbaz is doing the WSRF rendering but is waiting for feedback from
Mario before he starts writing the document (Oops!).
There was a discussion about whether the modifications could be made
to the existing documents but Mario and Shahbaz that it was probably
not.
Mario to point folks to the plain WS rendering.
A discussion on DTF and DTI followed and the rendering of it. David
wants to ensure that messages to be exchanged with the DTF and the
DTI are the same. DataMinx are using a RESTful approach. Shahbaz and
Mario had concerned about being able to do interoperability between
a RESTful implementaiton and Web Services based one. Maybe what the
DataMinx folks would like to do is a RESTFUL rendering of DMI...
Suggested approach:
- Mario will put in a document stub that the group will develop
over the next month and half. We'll have another telcon round
about then and evaluate whether there is enough enthusiasm/will/
need to continue.
If there is mileage in this we shall have to re-charter.
+-----------------------------------------------------------------------+
|Mario Antonioletti:EPCC,JCMB,The King's Buildings,Edinburgh EH9 3JZ. |
|Tel:0131 650 5141|mario@epcc.ed.ac.uk|http://www.epcc.ed.ac.uk/~mario/ |
+-----------------------------------------------------------------------+
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
