A gmake all of this package will build a binary, L3TrackAnalyze_x, which can be 
run with typical framework options as in e.g. runall

#!/bin/sh
./L3TrackAnalyze_x -rcp l3ftrack_analysis/rcp/runL3TrackAna_x.rcp \
  -input_file listfile:allRun187176.list

The file allRun187176.list contains a list of the data files I used.

There is a certain amount of setting up you need to do in order to run this 
code:

1. After setting up your release, setup D0RunII p16.01.00 SRT_QUAL=maxopt, you 
need to copy some configuration files into the directory in which you run.
Execute the script  $SRT_PUBLIC_CONTEXT/tsim_l3/bin/initScriptRunner.sh 
(You should only need to do this once).

2. You need to make sure you have a correct version of l3calib.ev for the data 
you are using.

3. Also, for post-shutdown data you should get l3cft_dead_fibres.dat which 
should be in the directory where you run.  This file comes from the CFT unpack 
tool package, l3fcftunpack.

4. You will need a suitable tools.lst file depending which tracker you want to
test.  I have examples for the GlobalTtracker, Pete's HTrack tool and my
CFT-only tracker, L3TCFTTrack.

5. If you want to test a new tracker you can simply add the relevant 
definitions to tools.rcp and make a suitable tools.lst with any parameters it 
needs.  You will also need to add an L3_TOOL_DECL line for it in my code at the 
top of l3ftrack_analysis/src/L3TrackAnalyze.cpp where the other tools are 
declared (and of course you will need to build the executable with a copy of 
you tool in the same release area).

That's about it for the first stage I think.  Running the executable should 
produce a root file, l3cfttrk.root, which contains a root tree of interesting 
variables.  I will probably add a few more things for the final version.  Also, 
I would like to be able to figure out how to get the timing statistics for the 
tools.  In the root file you have for each event all the track parameters and 
errors for the offline tracks (from d0reco), transformed into Level3 form for 
ease of comparison, and the track parameters and errors (which may be 
meaningless) for the Level3 tracks.  The Level3 tracks are generated by running 
the tracker within the executable, so the input data files need to be output 
from reco which has the raw data chunk retained.  The same analysis should work 
either with real data or Monte Carlo data that has been through d0reco, 
although I have only successfully tested it using real data so far.  I may have 
some problem with MC files to sort out.

Phase 2: the actual analysis.

Once you have generated the root file the analysis is very fast, using a root 
macro analysis.C and analysis.h .  This will be supplied in a standard version 
in the macros directory of the package. 

Compiling the macro properly is highly recommended.  It runs MUCH faster that 
way, i.e. seconds rather than minutes.

In root do:

.L analysis.C+
analysis a
a.Loop()

Make sure you also have the beamspot.dat file as the macro adjusts the DCA 
findability cuts to take account of the actual beamspot position.

This will run with the default cuts and produce 2 pages of standard plots.
Mnay other histograms also exist and can be examined.  A few more standard 
plots are in the process of being added.  The macro does track matching based 
on the axial track parameters only by default, but using the stereo is an 
option.  Look at the methods Options, Params, Bins, Findability to see what you 
can change.