CWNFramework: Column-wise ntuples in suez

CWNFramework is a package designed to make creating column-wise ntuples easy in both HBOOK and ROOT. The variables of the ntuple are specified in a separate text file; the ntuple structures and the interface with HBOOK/ROOT are taken care of automatically.

Here's a quick-start guide on how to use the framework. Assume that your processor is called MyProc, and you want your tuple class to be MyTuple.

• Check out the module CWNFramework from the CLEO CVS.
• In the MyProc/MyProc directory, create the file TUPLE_DEFINITION. The format will be documented later, but you can follow the example in HadronicDNtupleProc/HadronicDNtupleProc/TUPLE_DEFINITION. Some hints:
  • DOC begins a comment, and can be anywhere.
  • The first non-DOC line must be NTUPLE shortname description.
  • All variables must be in blocks. Blocks are started with the BLOCK blockname [OPTIONAL] keyword, and are ended either by another BLOCK statement or the end of the file. Blocks cannot be empty.
  • Array variables must be "indexed" by another variable; this specifies how large the array can be. Index variables are created with the statement INDEX varname min max; min and max specify the range of the index variable and must be nonnegative.
  • Integer variables are created with the statement INTEGER varname [indexvar] and floating-point variables with the statement FLOAT varname [indexvar]. The optional indexvar argument specifies that varname is an array indexed by the variable indexvar.
• From the MyProc/MyProc directory, run the script CWNFramework/Test/genDefinitions.py . This will autogenerate the header files for the tuple.
• From the MyProc directory, run the script CWNFramework/Test/genSkels.py . This will ask you a couple of questions and ask you to specify the class name of the tuple class you want to create. It will then generate some C++ files which specify the tuple; in this example they will be MyProc/Class/MyTuple.cc and MyProc/MyProc/MyTuple.h .
• You are now ready to use the tuple in your processor. An example of such use can be seen in HadronicDNtupleProc/Class/HadronicDNtupleProc.cc, although this file is rather long and involved. General principles:
  • Create an instance of either RootCWNtupler or HbookCWNtupler depending on which type you wish to produce.
  • Create an instance of your tuple class (e.g. MyTuple). You want to use only one instance of your tuple during your processor run.
  • Before you process any data (e.g. in your init() function), initialize the ntupler helper using the CWNtupler::init(string filename, CWNTuple* tuple) function.
  • Each time before you fill your tuple, call the clear() function of the tuple to set everything to zero.
  • The tuple variables are accessible as data members of the tuple class. (The array variables are just array data members.)
  • To fill the ntuple with the current entry, call CWNtupler::fill().
  • Before your processor quits (e.g. in your terminate() function), call CWNtupler::finalize().
  • In your Makefile, alter the line USER_LFLAGS := $(XTRA_LFLAGS) to USER_LFLAGS := $(XTRA_LFLAGS) $(shell exec $(C3_OTHER)/Root/bin/root-config --libs)
  • That's it!