Date:
October 14th 2004
Attendee:
David Alexander, Levente Hajdu, Efstratios Efstathiadis
, Jerome Lauret
Intro
discussion
Reviewed what we worked at the last
meeting.
Naming
-
We decided to come back to the U-JDL
naming (AHJDL seems confusing).
-
Composite Job was the latest agreed
upon name for a 'request'
-
Discussed of glossary and went
toward writing the schema
Andy's feedback:
-
publishing exec and leaving out the
low level JDL
-
simple request at Jlab: run over all
g1a data this code (that is all)
-
The J-Lab community adds a layer on
top of the J-Lab JDL designed to be as generic as possible without
an experiment language specific
-
As soon as one moves away from
files, datasets requires meta-data and one start speaking in
experiment specific language.
-
In principle, the RDL is fine but
since experiments are adding their own layers, it is not necessarily
a benefit to allocate manpower to change the current Jlab JDL to the
RDL.
-
Database tie up (some users may have
they private database)
-
Mark Ito [J-Lab, U-JDL and GUI ,
project up and down]
-
Basic things to get it to work
-
Memory the job will need
-
OS architecture
-
How long the job will run for
(normalized to CPU speed)
-
Disk Space (user care of the kind
of disk space). Only how much.
-
Instead of the names, tell where it
goes in terms of Work flow and this can be translated into where it
goes into Auger lingo.