> Dear ROOT developers, > > do you have plans to support C++ exception? Hi, While I would be wholeheartedly in favor of full support of C++ exceptions, let me point out that the ROOT classes are nowhere near exception safe. That is to say, exceptions in user code called via virtual functions or otherwise, within ROOT calls, are likely to propagate upward through the ROOT source and leave the system in an undefined, possibly unstable state. This is not a good situation to be in. Making ROOT and CINT exception safe would be, if I am any judge, a monumental effort. Making it easy for ROOT users to be or become exception safe would be an equally monumental effort, since if typical user code that I have seen is any indication, it is in this regard as bad as or worse the ROOT code itself. Just for starters, you have to support auto_ptr or a similar smart pointer, everywhere in the code; and auto_ptr is not in the gcc-2.91.66 library (it is in 2.95.2, but that is not used everywhere). So while it is on my wish list for ROOT, I wouldn't hold my breath. Personally I think more and better template support, especially of STL algorithms, is a higher priority (and probably no more difficult than exception safety), after const correctness. George Heintzelman gah@bnl.gov
This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:37 MET