> > Cool to know that journaling isn't such a big deal (in term of code > > complexity). 1000 lines of code. I'm kinda surprised ... ;-) > > Not only that. OpenBFS will be completelly working *WITHOUT* any > journaling at all. We have a transaction class that will be > responsible > for it but right now all it does is write the data directly to disk. > When everything else is working ans tested, we will add the > journaling > code. I may add that if anything went wrong now (i.e. running out of disk space - it doesn't have to be a crash), the file system may be in a strange state with defect inodes around and stuff - those problems will go away with journaling. Adios... Axel.