I completely agree about Linux and most OSS projects being uninterested in backward compatibility. Haiku has a far better model, I think, in trying to be backward compatible(bc). I think that the biggest issue is going to be bloat and concern about how bc holds it back. You don't want to find yourself in a place where it is too hard/impossible to upgrade the OS because you are required to be bc - that leads to developer frustration. On the other hand, what you described so well is user frustration. My recollection of the general belief is that bc should be maintained unless really, really, necessary. If there is some really compelling reason and there is just no way to keep bc, well, then, it has to be done. Otherwise, bc for me. :-) With the hybrid gcc2.95/4 builds that have been done, I don't foresee the need to break bc for a long time, if ever. I can foresee making some of the libraries an option to build/download (i.e. there are a lot of people who don't have R5 software, so they don't need the 2.95 libraries), but with all of the things that can be done to reduce this issue (making new classes, making new libraries, conditional loading, etc), I would say that you have found a new home. :-D Michael