Please can you check, if the behaviour is the same with g++-3.3? Sean Perry writes: > Package: g++-3.2 > Version: 1:3.2.3-0pre7 > Severity: normal >
> Now obviously, dynamic_cast requires rtti so using -fno-rtti is > silly. I ran into this on a C++ project where we had been using > -fno-rtti because rtti was not used. A recent commit added > dynamic_cast code and started segfaulting. It was believed that the > dynamic_cast was at fault and this caused some finger pointing. > Then the no-rtti was noticed. > > The compiler should give a warning (or perhaps even an error) if > code requiring rtti is used while the -fno-rtti option is in effect.