Development workflow¶
The instructions given in the setup section simply load all the compiled software into the current environment by updating the relevant environment variables, such as PATH
, LD_LIBRARY_PATH
, etc., such that files of all the executables and shared libraries can be located.
Regardless of whether you are running the installation from CVMFS or from a Docker/Apptainer container, the compiled files cannot be modified in a persistent manner. Thus to modify any package a new version of the corresponding code has to be built in a separate folder, where the user has the write permissions. After that the corresponding environment variables have to be updated to point to the newly built package.
General code development¶
In general, the whole software stack is built using the Spack package manager, as described in the corresponding mucoll-spack repository. This method allows to rebuild any piece of the software stack from the ground up, but is quite time-consuming, since it requires multiple preparatory steps:
- installation of Spack;
- setup of the code repositories;
- configuration of the development environment;
- rebuilding of the modified package;
- rebuilding of all the packages that depend on it.
The most frequent need for code development is limited to modifying an existing Marlin processor or adding a new one. In this case a faster approach is available.
Creating a Marlin Processor¶
A template Marlin package with a skeleton processor is available at TemplatePackage. It contains the following:
- A template structure for organizing your code.
- A
CMakeLists.txt
for compilation. - A skeleton processor that you can modify.
Note that official Marlin packages follow a slightly different template. This is because they are based on old CMake conventions. The template documented here has been updated for modern CMake.
To manage multiple packages, a simple workspace template is available at TemplateWorkspace.
More information is contained in the README.md
file of the repository, but in a nutshell allows to automatically recognize packages clones (preferably as git submodule) inside the packages
folder, providws a simple top CMakeLists.txt
file to compile them all and a setup.sh
script to modify automatically the MARLIN_DLL
environmental variable correctly.
Compiling¶
Start by making a fork or directly cloning the template repository.
The package can be compiled by loading the Muon Collider software environment and compiling as any CMake package.
Running¶
The software environment needs to be informed of the location of the library containing your processor. It loads all libraries specified by the MARLIN_DLL
environmental variable.
The following command will update the variable and call Marlin with the provided example steering file.
export MARLIN_DLL=$(realpath build/libTemplatePackage.so):${MARLIN_DLL}
k4run TemplatePackage/share/example.py --LcioEvent.Files output_reco.slcio
Modifying¶
The processor is implemented inside the src/TemplateProcessor.cxx
, with the header located in TemplatePackage/TemplateProcessor.hxx
.
The class implements the following functions:
init()
: Called at the start of Marlin execution.processRunHeader()
: Called at start of a file.processEvent()
: Called for each event.end()
: Called at the end.
Parameters are declared inside the class constructor using the following template:
registerProcessorParameter("MinPt", // Name of the parameter
"Minimum particle pT.", // Description
_minPt, // Variable that will contain the parameter
_minPt); // Default value
Parameters that correspond to input and output collections are defined slightly differently.
To define a parameter containing an input collection name:
registerInputCollection( LCIO::MCPARTICLE,
"InputCollectionName" ,
"Name of an input collection.",
_inputCollectionName,
_inputCollectionName
);
To define a parameter containing an output collection name: