Thanks. I read through the first one, and am reading the second one and this jumps out at me:
Quote:
My own order of preference is b) c) a). The fact that one filesystem will
offer features which other filesystems do not and cannot offer makes me
queasy for some reason.
|
Hmm, ext2/ext3 supports the immutable bit, file compression, and other features that chattr can set, but the other filesystems don't. Based on Andrew Morton's logic, ext(2/3) ought to be dropped from the Linux kernel, right? After all, it supports extensions which other Linux filesystems do not. Also, XFS, UFS, and ReiserFS all allow file sizes >2GB, but ext2 does not, therefore that is all the more reason that ext(2/3) ought to be dropped from the kernel.
Or, should ReiserFS extensions be allowed, providing that backwards compatibility is not broken?
Personally, I'd love to see more atomic file permissions implemented. -RWXRWXRWX is not as flexible as permissions allowed on NTFS, Netware, or even VMS filesystems (it's great to be able to deny a single user access to a filesystem resource should the need arise - without having to create a whole new group to do it). What is wrong with expanding the feature set of a filesystem providing that backwards compatibility is taken into account, or at least a mechanism for gracefully handling an incompatibility (exception handling, anyone?) - Andrew Morton's take on it is potentially holding back progress.
Personally, I *heart* ResierFS. I like the idea of
zero slack. I
need and use the >2GB support. I like the journaling (it saved my behind once as I cited above - a day's worth of work), and I like the performance. When I first read about btree-based filesystems (Longhorn) I was apprehensive, but when I came back to Linux last year, I read about ReiserFS3 and I decided to chance it - and I was pleasantly surprised that it performed just as well as, if not better than conventional filesystems. Then, it took a few months, but I finally ran into an issue where I NEEDED the journaling, so I rebooted, ran fsck, it replayed the journal, and I had my day's work back. I wish ReiserFS supported some of EXT(2/3)'s proprietary extensions, but all the filesystems are a work in progress and none of the various filesystems' extensions should be barred from the kernel - providing that extraneous/tangental utilities either downgrade gracefully or otherwise handle the missing or extra features well.
--Kim