Subject: Re: special-casing access to image files
To: None <tech-userlevel@netbsd.org>
From: Greg A. Woods <woods@weird.com>
List: tech-userlevel
Date: 08/19/2001 12:14:51
[ On Sunday, August 19, 2001 at 17:07:34 (+0200), Hubert Feyrer wrote: ]
> Subject: special-casing access to image files
>
> I see this -F added to ~all filesystem utilities. Is that really
> necessary? IIRC there's the Unix philosophy of "everything is a file", so
> why is there a difference when giving a filename over /dev/somedisk?
> Can't the fools be at least made auto-detect if th eunderlying thing is a
> disk image over a real disk?

The new logic in, for example, fsck_ffs is rather bizzare.  The only
thing done by '-F' seems to be to change whether or not you get asked
(when it's running interactively) whether or not to continue when a file
of the "wrong" type is specified.

This does seem to be a little more "motherly" than most Unix utilities.

It's one thing to require a device file when that's the only thing that
can possibly work for whatever reason(s), but quite another to make
distinctions that are no longer relevant.

If '-F' is to remain, at least in fsck_ffs, it might be nice if it would
also short-circuit the logic to use ioctl(DIOCGDINFO) when the first
superblock is borked.  I've not tried it yet, but I suspect the new
logic won't even work since the ioctl() failure will probably just cause
fsck to errx(EEXIT) out [[I don't see how the name of the device can
ever be NULL in the way getdisklabel() is currently called]].

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>     <woods@robohack.ca>
Planix, Inc. <woods@planix.com>;   Secrets of the Weird <woods@weird.com>