Even though I’m most fascinated by different programming languages, filesystems might be a close second. Kevin Bowling wrote a great piece on different filesystems in Linux and covered the basics like ext2-3, ReiserFS-3, JFS and the more modern ones like XFS (my current fav). Still, the most interesting is his discussion of “nextgen” filesystems like ext4, Reiser4, btrfs and some craziness like tux3. They all seem really cool but I have a hard time imagining any of them overtaking ZFS, the glamorized god-child of filesystems which seems to be about the only thing keeping OpenSolaris alive. Yes it’s hyped, but if it could get incorporated into the kernel (no, FUSE doesn’t count), it would be a fantastic development. I really wonder if anything can surpass ZFS in light of it’s rockin feature set. Specifically, the whole concept of storage pools is a huge selling point. My understanding is that filesystems can be allocated on top of logical disks called zpools. Now, it would seem that, much like the Linux logical volume manager (LVM)), physical disks can be grouped together in logical ones. The key difference is that with ZFS, zpools can be created and modified on the fly as first-class ZFS citizens with all the sides and toppings. I know that I’ve sung ZFS’ praises before, but I can’t overemphasize the advantages that the ability to dynamically add disks like this would bring to Linux. Still, with all of the different front-ends to GCC available, I wonder why non-C languages aren’t used. Ada) seems like it would be *so* much more testable, maintainable and stable.