For now, each time a command is executed by the Make process, a new
client process is created. Thus, it is very difficult to keep tracks of several important
things between compiler command calls, like configuration parameters,
matches between compiler support classes and compiler paths for
the IP adress a newly elected scheduler for example.
It should also increase performance because only tiny C programs will be launched each time a compiler command line is executed by the Make process, instead of a heavy Python interpreter.