Dear L2 Upgrade Interested Parties,

We thank the L2 upgrade group for their comprehensive talks, and applaud
the excellent condition of this project.  We believe the schedule for
integration in March 2005 is plausible and necessary, so that this
upgrade is smoothly operational before the numerous trigger upgrades
planned for the August 8 2005 shutdown are underway.  The L2 upgrade
is expected to improve the CDF physics capability and should therefore
be implemented as soon as possible, in preparation for the anticipated
luminosity increases.  Manpower appears to be adequate, but is dwindling
just at the moment when a push is required, and we strongly urge the
contributing institutions to supply additional manpower.  We appreciate
that a great deal of effort has gone into making this complex system
fairly straightforward, and note that this significantly simplifies
the work required for the new people to join.  Nevertheless, an overlap
period for training new people would be optimal.  We did not observe
any significant problems, and report below the requests and recommendations
we feel will assist you to make the system ready for standard operation
on schedule.  We tentatively schedule a meeting to discuss updates to these
issues for Wednesday January 26.  In late February, the committee will be
requested to sign off on project integration.

The project integration is coming along well:
o) Communication with the Run Control appears understood.
o) The diagnostic tools appear to be in good shape.  We remind you to
   make the diagnostic tools flexible enough to absorb future changes,
   and to prepare sufficient tools for expert debugging online.  In
   addition, we urge earlier integration into online TrigMon than the
   schedule showed (Nov.).  Bill Badgett (online DB) and Petar
   Maksimovic (offline DB) have agreed to be (and should be) consulted
   about database accesses.
o) Consult with the L2 CAL experts ASAP to solve the problem of wrong
   CLIST/ISOLIST protocol upstream if at all possible, before spending
   any further effort on compensation for this problem.
o) The active splitters for RECES should be completed as soon as possible.
   We find the design and 25% spares situation to be adequate, noting only
   that they must be maintained long-term, as the parallel structure of the
   system must be maintained for future upgrades.
o) The software spec is to reproduce the existing Run 2A system L2 software.
o) Additional manpower should be committed to the software development;
   we are concerned that the sheer volume of work needed to
   be done could cause a delay in integration.
o) Provide a specific plan for including the L2 upgrade system in the
   trigger table build procedure and online running.  For operation,
   JDL suggests to 1) invent a scheme to test a trigger-table shared
   library on a disk file on the build node, and 2) to develop L2
   integrity checks to be performed in a L3 filter.
o) Formulate soon the system configurations for "standard" vs.
   "diagnostic" readout.
o) Please do not expend significant effort on upgrades, e.g., FILAR, before
   the base project is operational.

The test runs taken with and without beam are useful and appropriate.
The additional beam time required for integration will be supported by
the Operations Group as necessary to improve future CDF physics capability.
However, every effort should be made to use beam tests wisely.  Beam tests
are to be negotiated with the Operations Manager; requests shall state how
much beamtime will be required, under which beam conditions, and whether
the recorded data are to be marked "good" or "bad."  No-beam tests of
clustering, including using the low-CAL-threshold trigger table (JDL) or
lowering Showermax thresholds may not have been sufficiently exploited.
The TL2D readout time should be measured during the shutdown.
How the the test stand and hot spares will be maintained after integration
should be carefully outlined before we meet again.

For operational support and maintenance, we have only good will at the
moment.  The MOU's covering all parts of the hardware/firmware must
be signed before the system can be considered operational; this process
is on track to happen in time.  The diagnostic tools are on track (note above).
Please discuss with the L1/L2 trigger spl's how to handle the
global L2 pager vs.individual L2 trigger subcomponent pagers.
The planned HW spares are adequate.  We have no reason to doubt the
reliability of the system which has been functioning
for a long time; the new active splitters should also be tested for
reliability.  We strongly stress the need for ongoing firmware support
as improvements become necessary down the road.  Automate the
firmware version control as much as possible to prevent accidents.
Provide a documented procedure for hardware configuration control
for Pulsars, daughter boards, firmware type and version, etc.

We will reconvene in a few months to define the actual switchover
to a trigger system with the L2 upgrade system driving the L2 decision.
At that point:
o) The Run 2B L2 system performance should have been shown
   to exceed the 2A system at any L1 accept rate and luminosity.
o) The trigger table algorithm and build procedure must be proven reliable.
o) Sufficient testing with the system in normal operation mode,
   i.e., without unnecessary diagnostics, should have been performed.
o) The online monitoring system must be finished, with histograms
   and straightforward criteria for CO checks defined.
o) MOU's demonstrating sufficient operational support for all parts
   of the hardware/firmware must have been signed.
Several weeks before the anticipated switchover, the TDWG heads will
declare a moratorium on new trigger code so we can make a smooth
switchover from 2A to 2B trigger algorithms.  (Parameters, prescales,
etc. can still change.)  At this point, with the blessing
of the IRR committee and the Operations Group, we would try running for
two weeks with the 2B system driving the L2 decision (or a typically
equivalent number of stores).  At the end of the two weeks,
if there are no significant problems, we consider the upgrade complete.
Trouble would reset the clock.

We are concerned about the long lists of tasks left on the software,
RECES and Cluster finder.  By October 15, please provide a detailed
prioritized checklist of jobs that need to be done for these items,
and give short progress reports at each meeting of the Trigger Hardware
Group.  No beyond-the-baseline tasks permitted in these reports!

Best wishes,
The L2 upgrade Installation Readiness Review Committee
(Bill Badgett, Roberto Carosi, Camille Ginsburg, Jonathan Lewis,
Pat Lukens, Steve Nahn, Jim Patrick, Kevin Pitts, Rob Roser,
Kirsten Tollefson, David Saltzberg, Michael Schmidt, Tom Wright)