Notes on Gen01 Customized Analyzer Engine
  =========================================

Distribution
------------
Here is the official home of the customized Csoft analysis replay engine.
A reference setup is contained in the subdirectory   replay
The runtime setup and the source tree are conatined in the respective 
TAR files, with a version identifier.
LINUX and SUN executables are also provided, with the same version ID.

To create a working setup, follow these steps:

A) at Jlab (quick and easy)
  1. change to the directory where you want to set up your replay
  2. copy the runtime directory structure
      cp -a /group/e93026/analyzer/replay ./replay
     or
      cp -dpR /group/e93026/analyzer/replay ./replay
  3. copy the latest exectuables ({pf} = 'SunOS'  or  'Linux')
      cp /group/e93026/analyzer/engine_{pf}.exe  ./replay
      cp /group/e93026/analyzer/synchfilter_{pf} ./replay
  4. edit ./replay/REPLAY.PARM to reflect the actual replay
     directory and the desired data source location

B) in general

  1. download the latest RUNTIME distribution TAR file, 
     gunzip and untar it in the desired location(!) with the command
      tar -xzf {filename.tar.gz}   
     or, in case this doesn't work (SUN?),
      gunzip {filename.tar.gz}   
     followed by 
      tar -xf {filename.tar}

  2. download the latest executable for your platform 
     (into your ./replay directory) or roll your own (see item 5)
     Currently, only SunOS and Linux are supported!!!

  3. OPTIONAL if you want to, or plan to compile your own executable,
     download (or copy) the Csoft distribution and place it in a 
     suitable directory -- gunzip/untar it as you did with RUNTIME.  
     Be sure to update the location in step 4!

  4. examine the files REPLAY.PARM  and  replayup_{platform}
     and adjust the paths and file locations as needed, then execute
      source replayup_{platform}

  5. to create your own executable, first download (copy) the latest
     SOURCE distribution and gunzip/untar it (inside ./replay)
     then change into the new SRC directory and execute
      make
     afterwards, move the resulting executables
      engine_{platform}.exe
     and
      synchfilter_{platform}
     into the ./replay directory
   


Setup and Runtime
-----------------
We have our own copy of the Csoft tree at this point.  This provides
a fixed set of libraries and we do not need to be concerned about 
changes in the public location.  Those are located in the group space, 
same as the other analyzer files, and are used by the compiler/linker.

On any Jlab CUE LINUX PC, simply unTARing the runtime files ought
to result in a usable analysis setup, though the declarations in the
setup script should be verified ("replayup" -- platform dependent). 
See the detailed instructions above. 

If the system is NOT a Jlab CUE machine, some paths may need to be
modified, particularly  CERN_ROOT and of course the location of Csoft.
Also, the location of the raw data files (replay input) will need to be
adjusted in  REPLAY.PARM


Raw (input) Files
-----------------
The raw files are the input to the analyzer engine.  Their location and naming
depends on the experiment, platform and status.  Here, we will briefly describe
the method used to acces the file in 2 cases:  "near-online" replay of a recent
run from the CDAQ buffer disk on one of the CDAQ machines, and the replay of a
run that' been written to tape on a Jlab CUE machine.

To use a run on the CDAQ buffer, simply define the source in REPLAY.PARM to be
  g_data_source_filename =
        '/net/cdaqfs/hallc/home/cdaq/gen/coda2/runlist/e93026_%d.log'
(without the line break!)  This will work as long as the data are on the buffer
disk(s).

To replay a run that's already written out to tape, the raw file must first be
read in and the data source in REPLAY.PARM must be set to match:
  g_data_source_filename = '/cache/mss/hallc/e93026/raw/e93026_%d.log'

To get the run from tape onto the cache disk, it must be requested via the
"jcache" command, e.g.
  jcache /w/mss/hallc/e93026/raw/e93026_22597.log
to access the raw data.  Note that this example will not work, as the run used 
is from Gen98 and the file actually has a different name (gen98_22597.log).