GlueX Software Meeting
May 24, 2005

Attendees: Paul E. Mark I. Craig B. Simon T. David L. Elliott W.
Zisis P. Mauricio Nikolay Matt B. Rafael H. Greg R.+ Ed B.

Agenda:

1. Calibration/Parameters Database (see below )    (Nikolay)

2. Meeting format (All) 

------------------------------------------------------------

We spent the entire meeting discussing databases as applied to
calibration constants. The discussion was initiated by Nikolay's 
e-mail questionaire which can be found here:

http://tantalus.phys.uregina.ca/drupal/gluex/?q=dbquestions

There was general agreement that we should avoid OO databases
and should use MySQL.

Greg made a few suggestions:

1. Make communication be mediated via XML. This relieves the
reconstruction software from needing any specific knowledge of
the database structure. It also makes it easier to switch the
input from a database (possibly remote) to flat files.

2. Access to the database should be done through web sevices.

3. Calibration data should be time stamped and never deleted
thus preserving history.

Mark reviewed some highlights of a talk he'd given at a previous
GlueX meeting(Oct. 26, 2004) relating the CLAS experience.
The CLAS database design used an index table which provided an
extra level of indirection to the constants. The index table
could then be copied to "freeze" constants or, to use as a
private copy of the database. One problem that emerged from
this functionality was that CLAS run groups would use the 
private index tables for analysis and only later (if at all) 
copy the final values to the master table. If possible, the
system needs to be designed to allow functionality of developing
constants in a private area, while discouraging data processing
from there.

The discussion went on for over an hour and there were more
things discussed. There are many potential pitfalls and design
considerations to pay attention to here so it is clear more
discussion on this topic is needed.