> In a perfect world, we'd have:
> 
> 	newfs_ffs
> 	newfs_lfs
> 	newfs_msdosfs
> 	newfs_adosfs
> 
> ...etc.  There would be a signle "driver" program, called "newfs", which
> would be invoked like:
> 
> 	newfs -t ffs /dev/rsd5f
> 	newfs -t msdosfs /dev/rfd0a
> 
> If no -t argument is given, the system would default to ffs.

Actually, i'd say, in a perfect world newfs would be a hard link to
the 'default file system type', which would currently be ffs though
might be something different if we switch to a different 'default'
file system format in the future.

while a 'switch' program is nice for symmetry with mount and fsck, it
doesn't make so much sense for newfs.  both mount and fsck _need_ a
top-level driver program to do the right things with file systems, and
invoke the appropriate lower-level programs.

newfs has no such need, and it's wasteful of space (disk space) and
time ("newfs -t foofs" is three characters longer than newfs_foofs
8-)...



cgd