Meeting on RunII Computing Issues 1/10/96

The third in a serious of meetings between the Computing Division's scientists 
working on CDF and D0, CDF and D0 Run II upgrade computing project management, 
and Computing Division management was held on January 10, 1995. 		

The Computing division attendees were :Mike Deisburg, Pushpa Bhat, Steve 
Wolbers, Stephan Lammel, Eric Wicklund, Irwin Gaines, Liz Buckley Geer, Rob 
Harris, Adam Para, Lee Leuking, Joel Butler, Paul LeBrun, Amber Boehmlein,
G.P. Yeh, Vicky White.   

At both the previous meeting and this meeting we were joined by Rob Plunkett, 
who is co-leading the CDF Computing upgrade effort with Liz Buckley Geer.

D0 have very recently identified their leadership for Computing for the Run II 
upgrade and both of these people - Bruce Gibbard and Wyatt Merrit were kind 
enough to attend the meeting, together with Stu Fuess.  They had not had time 
yet to gather their thoughts and so deferred a presentation of their ideas for 
how to organize the work ahead for Run II to the next meeting.

The meeting focussed on trying to identify possible areas where a joint 
working group of D0 people, CDF people,  (both Fermi based and University 
based) and Computing division people could work together to do any one of the 
following:
characterize needs in common between CDF and D0, identify potential tools, 
make recommendations on standards or packages to adopt, specify projects to be 
carried out largely by Computing professionals, share and discuss ideas on how 
to proceed, etc.  

It was early on agreed that the list did not represent any priority ordering 
of issues needing addressing - that needs to be done now in consultation with 
each experiment to identify those working groups most urgently needed and 
standing the greatest chance of success. 

The following potential working groups were identified:

1) Editors - standardizing on an editor and providing templates and additional 
features to help the collaboration is a D0 goal.  They had recently chosen 
EMACS. CDF feels less strongly about a standard editor, but might have no 
objection to one, along with the value-added things.

2) Code Management - everyone clearly needs this, and in the broadest sense of 
the word, not just version control.  D0 has a working group currently active.  
The Computing division had a working group active for some time, with CDF and 
D0 participation (but not quite the right time to get their attention).  They 
came up with CVS as a base tool, and wrote a report which nicely identified 
the broader issues of code management - including distribution and validation. 
It was hoped that all working groups considering this issue would take this 
report at least as a starting point.

3) Interactive graphics - for display facilities.  Can some core tools be 
agreed upon such that the experiment specific parts can be layered on top.  

4) Software engineering tools, design tools, validation tools.   This clearly 
is a large area coupled to methodology choices.  The computing division has 
several projects with expertise in specific methodologies and tools and 
certainly can arrange some presentations to share information and start the 
discussion.

5) Languages, compilers, what model of multi-language support (if any), 
provision of useful and standard class libraries (assuming C++ is somewhere in 
the picture), standard template library role.  There were lots of issues which 
seemed to fall in this area, the not least of which was sorting out which 
issues really should be a part of such a working groups charge and 
consideration.

6) Data management, data organization and data access.    Everyone agreed that 
this was a core issue.  The term data model was also bandied about in this 
context -but clearly not everyone was using the term in the same way and with 
the same basic meaning.   This probably needs sorting out in order to decide 
if there is some other working group more focussed on distributed versus 
centralized model of computing,  or some other issue related to some people's 
use of the term data model.

7) Analysis Tools - what will be the equivalent of PAW and PIAF for the future 
and what should we be doing for this.  What about replacement of CERNLIB in 
general?

8) Production Tools - the next generation CPS like tools - or what? for 
managing production and production systems.

9) Databases - what, where to use

10) Simulation - is this a area of commonality between CDF and D0 or too 
different to merit joint work?  Work is going on in the Computing division on 
collaborating with CERN on GEANT4 and on working on MCfast - originally for B 
experiment simulation but now shown to be more generally applicable.  



It was agreed to try to meet every other week and tentatively Thursdays at 
12.30 to 2.00 was chosen as a regular meeting time *** except that the next 
meeting will probably be in three weeks on February 1. ***

Participants were encouraged to discuss these potential working group areas 
with others in their experiments, refine their ideas and come back with 
suggestions on which working group's should be actually launched and charged.

Hope I didn't leave some key idea of someone's out, apologies for misspelt 
names.
			Vicky