The Software challenges for the next decade

Rene BRUN
CERN

The past 12 years have seen a considerable evolution of the software systems, libraries and
tools in HEP. We have painfully moved from Fortran to  object-oriented languages, tried
commercial solutions including object-oriented data bases and graphics systems. It took many
years to develop the large systems like ROOT or Geant4. We are facing interesting challenges
in the coming years with the new possibilities offered by parallel architectures, from
multi-core processors to large clusters on the Grid(s). This will imply drastic
changes in the way we build, distribute, install and use software.
I will address some of these issues in taking as an example our evolution model for the ROOT
system.