Dear Sirs, I have a request about the notion of G4Exception. It is not real exception, it is just function (not a class!) which calls abort(). I am not sure that we would be pleasured by such behaviour in the nearest future. For us it would be significantly more prefferrable if G4Exception would throw "true" exception (it is very convinient for us if this exception would inherit from std::exception class), and the "client" (GiGa or Gaudi) is responsible to catch this exception and perform some action. With my best personal regards, Vanya Belyaev
By design G4Exception is a simple function printing some diagnostics and aborting the application when invoked. For reasons mainly related to performance, C++ exception handling is not used in Geant4. However, we recently extended the current mechanism by introducing the capability of handling/customising G4Exceptions by making G4UIManager "state" dependent, therefore allowing users to define the desired application behavior once in 'abort' state, also allowing to print -any- kind of diagnostics or performing 'safe' operations before the actual abortion. The user documentation needs to be updated to reflect these features. Handed this over to Makoto... :)
Documents should be added by next public release...
Now the user has an UI command (/control/suppressAbortion) to suppress the G4Exception at event loop. Of course messages are still displayed. It is contained in Geant4-V03-02-ref-01.
