I tried to look in the documentation and some examples, but what I found is:
The Process method of TProof takes as parameter a string=the name of the file where my TSelector is implemented.
What I would like to do is:
in the main() function of my program to create an TSelector object; call it’s Init() method etc. Then to create an TProof object and call it’s Process method giving it the pointer to the TSelector object.
and then I would compile the whole thing: (something like):
g++ -O0 -g3 $(root-config --glibs --cflags) MySelector.cxx simple_ana.C -o simple_ana
and run it.
But seams to me that such approach is not possible since the Process method of TProof doesn’t take as an parameter a TSelector object.
Did I totally misunderstood something and the way I want to do things is a nonsense?
This functionality is not implemented, basically because you need anyhow to send the selector code on the workers to be able to understand the TSelector object that you would send from the client.
What do you need to do in Init() that cannot be done in SlaveBegin() based , for example, on the input from the InputList ?