Re: rmdir system call

Ulrich Windl (Ulrich.Windl@rz.uni-regensburg.de)
Thu, 21 Mar 1996 09:42:19 +0100


On 20 Mar 96 at 23:18, Leonard N. Zubkoff wrote:

> Date: Thu, 21 Mar 1996 08:35:10 +0200 (EET)
> From: Linus Torvalds <torvalds@cs.helsinki.fi>
>
> Personally, I think that the initial // behaviour is totally broken,
> because it's against the whole unix philosophy of having a unified file
> system name space, but we don't really need to care about it, and linux
> handles that the "correct" way (as do just about all the other "normal"
> unixes out there).
>
> Actually, I think the standard Unix behavior is broken in this regard, since it
> doesn't handle a networked environment uniformly, and NFS with its automounting
> is an abomination. I spent a number of years using Domain/OS, in which //
> referred to the network root directory. With Domain/OS, you could construct a
> file name which was valid anywhere in the entire network and referred to the
> same object, such as //atlantis/lnz/foo. That is there was a single namespace

Maybe /../atlantis/lnz/foo would have been better (Back in '85 I saw
some Edition7 having that extension).

> across the entire network, and you didn't need to "mount" every file system you
> might ever want to access. File names beginning with a single / were relative
> to the local node only.
>
> I know that some systems make the Unix naming a little more palatable by
> reserving /net/hostname/filesystem for referring uniformly to network files,
> but that's really no cleaner than // in that it makes /net a special piece of
> syntax. Personally, I always found the //nodename prefix a reasonably elegant
> solution.
>
> Leonard
(Ever tried a findfirst("C:\\.", ...) in MS-DOS? ENOENT!)