Janito Ferreira Filho <jvffprog@xxxxxxxxxxx> wrote: > I'm currently leaning toward implementing something along the > ext2/3/4 path. > Maybe NILFS I could implement later, outside of GSOC. Anyway, maybe I > might > not get as far as ext4, but either way, I am aiming for some write > support, > whichever I do. As far as I understood, there isn't write support for > ext2 yet, is > that correct? If so, I think that would be a nice place to start. Any > opinions? That sounds like the way to go. ext2 write support shouldn't be too hard, although I would try to go for ext3 already, initially (we have ext3 read support, too, as long as it was cleanly unmounted), as this should give you a better picture of how to implement it properly. Another option would be to start with adding read support for ext4 - this would give you a complete overview over ext2/3/4 functionality and inner workings, and might actually help with implementing write support, too. In any case, the existing ext2 code should be a good base to start from. Alternatively, one could try to port the Linux code into a new file system; I'm not sure that would be much easier, though, and having an MIT licensed fully implemented ext4 file system in C++ sounds appealing to me as well. Bye, Axel.