> > Is somebody still actively working on LFS?
> >
> > Would it be possible to import ReiserFS into
> > the NetBSD Kernel?
>
>  Reiser FS has few issues that make it highly controversial file system. Yes,

and highly commercial and highly buggy filesystem.
i've used it with linux >6 months and then wasted 3 days to rescue data as
it made kernel panic when entering defective directory (copied 1 by 1
dir to ext2, sync every time and reboot every each crash).
it's NOT FUNNY.
now in reiserfs such nonsenses are done like "squid optimized filesystem"
(why don't implement it in userlevel in squid itself, and use raw
partition?!) and other good sounding commercially things.

>  it can be probably very (or even extremely) fast on some types of workload
>  but lack of reliable repairing tool like `fsck_reiserfs' makes it pointless

fsck_reiserfs should be linked to mkreiserfs as in fact mkreiserfs is the
only thing to do after fsck_reiserfs ;)

>  and very dangerous in real use. What they currently have under the name
>  `fsck.reiserfs' is more like a stub and, moreover, the kernel code refuses
>  to mount the dirty file system even in read only mode. In case if you have
>  bad disk block in metadata area on Reiser fs populated volume, it is
>  completely impossible to do anything with your data (you can't even mount to
>  make a partial backup of your file system) and defective hard drives still
>  do appear (recent story with IBM IDE DTLA 75Gb disks, for example).

don't know what's now but recently reiserfs had no support for bad blocks
or even I/O errors. any I/O error = kernel panic.

reiserfs was an excellent idea, but that's all.