Minutes of DAQ/Online Calibration Meeting 01/07/00

Agenda:
=======
Minutes of last meeting     Rick St. Denis  5
(see below for corrections so far)
Follow up from last meeting    
News


Round Table (5 minutes each) - names not confirmed.
-----------
COT/Muon:                  Ashutosh Kotwal
SVX:                       Steve Nahn
Database:                  Jack Cranshaw
Calorimeter/DBana:         Vaia Papadimitriou
CLC:                       Alexei Safonov  
Consumer:                  Kaori Maeshima
Access Methods:            Rick Vidal
ShowerMax:                 Steve Kuhlmann (Email)

Presentations
-------------
Validation and testing     Rick St. Denis 10
Templates                  Jack Cranshaw  10
Joblists update            Rick St. Denis  5

Discussion
----------
Next meetings
Workshop
Milestones

AOB

Corrections to minutes of 10-dec-99


Minutes of the DAQ/Online Calibration Meeting on Jan 7, 2000

Minutes of last meeting     Rick St. Denis  5
(see corrections above)
Follow up from last meeting    
News

- ROOT may be used both as a Calibration Consumer in line with the
  general consumer framework and with the offline.  The production jobs
  in offline will not have any histogramming. This will be put into a
  Run Quality job which looks at the data as soon as the reconstruction is 
  done. 

- Any work done in EDM1 format will be converted to EDM2 by Rich Glosson
  Note: in another meeting later in the day, it was decided that all such
        requests should be given to Rick St. Denis (calibration) and 
        Jim Patrick (other online).  The requests will be processed according
        the priority for achieving milestones.  Rich Glosson will do the 
        conversion and explain what he has done if you are interested.

- The Consumer Server Logger will be installed on all machines serviced
  by the CDF code distribution system.

- Apologies (especially to Alexi) for the confusion over the start time.

Round Table (5 minutes each)
-----------
COT/Muon:                  Bill Orejudos
========================================
Bill read a file of data, sent it through the Consumer Server Logger
to his consumer and wrote it to the database.  He is close to getting
this done in real time.

SVX:                       Steve Nahn/Rick St.Denis
===================================================
There was confusion about the meaning of the keyword on the Calibration 
database and how to relate it to the hardware.  This has been sorted out
and another test is scheduled for this afternoon.  

Pedestals have been read from the Database and downloaded, resulting in a 
pedestal subtracted signal of zero; however, the loop of reading, computing
and saving pedestals in real time and writing to the database has not been 
done.

Rick Vidal commented that a real configuration of the database is needed with 
the final hardware positions marked and only those pieces available marked 
online.

Database:                  Jack Cranshaw
=========================================
Views of the database have been created for the calorimetry.  For exmaple,
the calorimeter view "view" contains information on:

eta, phi, PMT number (chanelnumber)
QIE number (subaddress)
ADMEM number (slot)
Crate information (CRATENAME, NODEVMEPROC)
Diracchannel (DIRACCHANNEL)
DIRAC CARD (DIRAC)
Wedge number (WEDGENUM)
Rack information

The CHANNELNUMBER has information as well that links the geometrical
location of channel with its electronics location.

Dennis Box needs to wite software to access this view.  In this way the 
calibration data can be linked to the hardware locations.

Rick Vidal brought up the question to ponder: What are the demands on the 
development version of the database.


Calorimeter/DBana:         Vaia Papadimitriou
=============================================
Vaia showed transparencies indicating that the gains of channels are 
relatively stable and that in the HCAL there is clearly a dead channel.
She described some problem with the read position of a plug hadron crate.
Rick St. Denis asked if the configuration has to be done from the text files
or whether it can be done from the Hardware database. Jim Patrick said that 
since the cards are now in their final locations, it makes sense to use the
hardware database for the calorimetry.  Jack pointed out that the plug was not
yet in the database; however, Jim said that this has been entered this week
although it needs to be checked.  Vaia agreed to talk to Bill Badgett about
how to use the hardware database to run the calorimeters.

Steve Nahn asked if the histogramming code done in the consumer program 
can also be used in DBANA.  Vaia pointed out that since the code is all 
C++ (or pseudo-C++) and can be placed in a macro supplied by the subdetector
group, this flexiblity exists.  

CLC:                       Alexei Safonov  
=========================================
Alexei said that the Trigger supervisor and Run Control were used to read out 
their crate on the first floor.  They (Nomerotski/Chelbana) have written 
D banks and looked at them. They found that if they read the COT and CLC 
together, there was only 100ps slewing in time.  This is very good news.  
They need extra voltage on the crate (this afternoon) to be able to read 
out the scintillator in the collision hall at the same time.

Alexei asked how the calibration can be down for LED given that they have to 
set a variety of steps.  Jim Patrick said this is forseen and Rick St. Denis
said that although he did not know, this is a common problem and needs to be
solved for a variety of systems.

Consumer:                  -
============================
No report

Access Methods:            Rick Vidal
=====================================
Dennis Box has been working on streams to access the Database.
Rick suggested we invite him to talk at the next meeting. The invitation
will be extended.

ShowerMax:                 Steve Kuhlmann (Email)
--------------------------------------------------
     The official version of Run_control can now calibrate and
filter channels in the central and plug strips.   It takes 
48 DAC settings over 8 ranges,  fits each range to a line,  
filters bad channels, creates and loads the lookup table.  All 
in the processor.   Takes 3 seconds per SMXR card in the 
central,  5 seconds in the plug.  


Presentations
-------------
Validation and testing     Rick St. Denis 10
============================================
Rick Described the new SRT feature

gmake test

and how to implement it. This can automatically be used in validation.
Details and directions may be found in   the
web page on this item.

Templates                  Jack Cranshaw  10
============================================
Jack described an idea for C++ templating of an AC++ module that
allows one to write a few lines of code and receive the data for each
channel in a method. The pedestal method needed in such a scheme only
requires a couple of lines of code to be written. 

Rick St. Denis pointed out the technical aspects of the implementation will be 
discussed with the gurus and a name has been found to implement the template.

It is hoped that nobody will have to work to 
- write a d bank pedestal job
- figure out how to handle storable objects
- figure out how to loop over edm2 objects

This also solves the question of whether one writes in EDM 1 or EDM2 since
the template can handle this -- the user need only know the format of the
key and information associated with each key (ie channel) and a second key
and its information (the calibration!).  

Jack's transparencies may be found in this
location

Joblists update            Rick St. Denis  5
============================================
No changes


Discussion
----------
Time ran out, but:
Next meetings -- Jan 21 at 9am, theater
Workshop -- Online Calibration Steering group will meet Jan 13
Milestones -- Rick will update the errors in his SVX milestones 
              He will discuss the COT/MUON milestones privately with
              Bill Orejudos before showing them at this meeting
              He will get milestones for CLC at the Steering group meeting.

AOB
===
Steve Nahn asked about sending messages to merlin from consumers and how the
consumers are started from Run Control in the first place. Jim Patrick answered
that the computing division had software that caused an rlogin and executed
the program.  In previous runs they required that the program be executed
with a specific run type and run number to ensure that the correct data were 
consumed.