user filesystems 1803

hi steve: I appreciate the links, but going down the list seems to only reinforce that the mainstream linux file system set has not done much for simple-user features lately:

cramfs are part of the linux kernel (hooray), but read-only filesystems (booh!). squashfs is read-only, too.

e2compr begins its website with a statement that it has a long and checkered history, and then says that a port from kernel 2.4 to 2.6 is planned.

fusecompress is the work of one individual in the czech republic. it doesn't really have any description or docs. I hope it will continue, but noone knows. very low user base for now. compFUSEd falls in the same category for now.

cloop---I didn't think of this. nice and maintained. I wonder if it could work...homework.

ziosofs---read-only.

ext3cow is a nice feature idea. maybe it will make it into the user base. I am not clear how it deletes old files when one runs out of storage space, which could probably happen fairly quickly.

clearcase---this seems to be some IBM enterprise product, not a file system.

undeletion and ext2fs: these are from 1999. in fact, someone had written an undel, but ext3 broke it. the linux journal article dates to kernel 2.0.x.

I had looked at many of these before I posted, which is also partly why I was dismayed. I would guess it would not be too controversial to say that progress on simple end-user-centric filesystems has been very slow. to me, and more controversially, progress in this respect but not others seems to have been in very mild reverse.

I vaguely recall that one of the commonly stated goals of the next few years is for linux to conquer the simple-user desktop. I would guess that an end-user oriented feature-rich filesystem could help.

regards,

ps: thanks, josef.