Norman,
Is there any documentation on the CEA -- Common Execution Architecture?
How does it differ from say CORBA (http://www.omg.org/library/wpjava.html)?
I assume that the difference is the addition of a parameter system?
Steve.
-----Original Message-----
From: Starlink development [mailto:[log in to unmask]] On Behalf Of
Norman Gray
Sent: 04 October 2004 11:40
To: [log in to unmask]
Subject: IVOA general notes
All,
Here are some further notes about the meeting last week, which aren't
specific to any of the individual sessions (some notes for which I've
already passed on).
- Introductory talk -- state of the IVOA, Andy Lawrence (IVOA chair,
this year)
- Mentioned VOPlot (from India), SEDPlot (presumably from IfA, but
he didn't say that), and Topcat (and mentioned it was from
Starlink!).
- Things coming up: Euro-VO: VOTECH project proposal funded (no
further details); AstroGrid's RC1 scheduled for the end of this
year.
- [Bob Hanisch included here a mention of the NVO summer school, at
Aspen 13-17 September this year: tutorials and student projects:
http://chart.stsci.edu/twiki/bin/view/Main/SummerSchool]
- Next InterOp meetings are May 2005, Kyoto, and then Autumn 2005,
ADASS+Interop, Madrid
- IVOA issues
- Data models
- VOTable vs. TabSky (latter is a registry thing -- `more XMLie'
(what!)) -- how does the IVOA evolve new standards?
- Proposed VOTransient WG -- should this be an Interest Group
instead?
- Grid compliant vs Web-compliant (have tended towards the latter
so far -- is that good?)
- Peter Quinn has proposed an IVOA `business plan'. discussion to
come
- It seems that the IVOA was singled out for praise in some
recent OECD report (which I think Andy was involved with, in
some capacity I don't know about).
- Progress
- Data interop -- good progress, VOTable, SIAP, etc
- Resource interop -- good progress, RSM progress, and of mutual
harvesting of databases
- data integration -- not there, SkyQuery is the nearest so far.
Need real semantics -- grammar rather than types; UCD++,
ontologies? (all the things I've been saying...)
- single signon -- limited IVOA attention. `Shibboleth' appears
to be gaining juggernaut status, but things have to be
compatible with local academic/university demands
- complex jobs and workflow -- no attention so far, but the
bioinformatics folk are well ahead on this, to the point where
they're talking about patenting such descriptions.
- In the final IVOA session (Wednesday morning), Bob gave a rather good
talk about the
future of the VO, and the
pressures and requirements. This should appear on the wiki, before
long, and might be worth looking at. Mentions Euro-VO: VOTECH
(technology centre) has now been funded -- should find out more about
this, I suppose.
- Guy at coffee on Friday
- AG have developed CEA -- Common Execution Architecture. Wraps
Java apps (typically, but not exclusively). Handles passing
data between apps, and
giving them access to local files corresponding to virtual
files. This is the integration with workflow. Presumably if
ADAM were integrated with CEA then the Starlink apps would be
CEA apps instantly. Guy claimed that AG have been banging on
about this for a year, and that this is the way that the apps
would be glued together.
- Their paradigm is that the applications are all on the servers
-- the services are the applications -- so that there is little
or no scope for standalone desktop applications (though Guy
suggested that Topcat was the sort of thing that would be an
exception). I think their only actual application is a wrapped
SExtractor, though.
- There seems to be no real distinction between a pipeline and a
workflow, except that a workflow is more dynamic in the sense
that a user will (in principle) plumb the various bits together
themself. Thus a pipeline, in this sense, is a workflow that
someone else has written for you and that is fairly static.
- Guy's architecture document is ready now, and about to be
available.
- MySpace about usable now or in a couple of weeks. John Taylor
is the person who's handling the code and release. Linked to
the MAVEN effort.
- Talking to Andy Lawrence about this later, he said that,
although each of the tools was on a server, there wouldn't be
anything to stop you writing a tool yourself (say a prototype
source extractor), running it on a private server locally, and
using that as a `remote' service in a private workflow.
--
----------------------------------------------------------------------
Norman Gray : Physics & Astronomy, Glasgow University, UK
http://www.astro.gla.ac.uk/users/norman/ : www.starlink.ac.uk
