Summary of meeting between Jon Hays, Moacyr Souza, Terry Wyatt and
Taka Yasuda to decide on modifications to the L3 monitor statistics
displayed by the online daq_monitor.    3rd June 2002.

We agreed on the basic principle that the daq_monitor should in the
main display the number of true calls and passes for the scripts and
filters.  At the moment the statistics are confused because statistics
are displayed for a mixture of ordinary events and Mark&Pass events.


Modifications to be made by Taka to the daq_monitor:
====================================================

In the SCRIPTS section the following information will be displayed:
number of calls, number of true (not M&P) passes, number of forced M&P.
This last quantity will be renamed "forced", because the current name
"unbiassed" is confusing.

A toggle will be provided to switch between displaying the NUMBER of
passed/forced events and the FRACTION of passed/forced events.

A toggle will be provided to switch between displaying the statistics
INTEGRATED since the beginning of the run and statistics for the LAST
MONITORING PERIOD only.

The number of L3 nodes contributing to the displayed statistics will
be given.


Modifications that require ScriptRunner changes (by Jon or Moacyr)
==================================================================

In order that the statistics given in the FILTERS section are valid
for true passes and not confused by M&P events, extra counters for each
filter of the number of calls and number of passes are needed that 
are incremented only if all previous filters in the relevant script
returned true passes.
(This requires a new flag to be set in ScriptRunner for each filter if
ANY of the previous filters were forced unbiassed.)

The end result of these changes will be that the number of passed
events for the SCRIPT will equal the number of passed events for the
final FILTER in the script.  This is something that is not true now
and causes no end of confusion.

In the cases where more than one L3 script hangs off of one single
L1/L2 bit, statistics will be kept for the logical .OR. of all the L3 
scripts.  This will allow the true L3 rejection and rate to tape for a
particular L1/L2 bit to be ascertained.  Because of the (often large)
degree of overlap between the L3 bits this information cannot be
ascertained at the moment.

 
Scribe: Terry.