www.delorie.com/archives/browse.cgi   search  
Mail Archives: cygwin-developers/1998/08/14/06:31:13

From: lhall AT rfk DOT com (Larry Hall)
Subject: Re: mounts, ls, and filename completion
14 Aug 1998 06:31:13 -0700 :
Message-ID: <3.0.5.32.19980814092230.009e93f0.cygnus.cygwin32.developers@pop.ma.ultranet.com>
References: <35d51ff8 DOT 27329769 AT mail DOT goodnet DOT com>
Mime-Version: 1.0
To: cygwin32-developers AT cygnus DOT com

At 11:34 AM 8/14/98 GMT, jeffdb wrote:
>Hi Geoff ;^)
>
>As was pointed out to me when I mistakenly
>said that // was just a kludge to allow a brain dead
>drive table system to emulate a continuous
>unix style FS, // is required by the posix
>standard.
>
>That doesn't mean you have to use it, but programs
>that claim to be posix must allow it, and attach no
>special significance to it.
>
>Since cygwin32 has to support the syntax anyway
>if it wants to claim posix compliance, I don't see much
>point in trying to come up with an alternative.
>
>Either all drives would have to be automatically mounted
>at dll startup, and no programmatic access to the mount
>table could be allowed, or you would have to come up
>with something similar to // that really would be an
>incompatible kludge.
>
>OTOH if you want to rewrite the // support so that
>it will accept both //computername/drive:/dir/file.ext
>and //drive/dir/file.ext, more power to you, I don't see how
>you could possibly support both, but you are the cygwin32
>GOD ;^)
>
>possibly a : in the path could force UNC?
>
>On Thu, 13 Aug 1998 05:00:21 +0000, you wrote:
>
>>On Mon, Aug 10, 1998 at 09:31:26AM -0400, Larry Hall wrote:
>>[...]
>>> This should be a FAQ (is it already?)  The mount utility has much the same
>>> semantics (although not the same syntax!) as UNIX variants.  It does not,
>>> however, have the requirement that the directory to which you are mounting
>>> must exist.  This is unfortunate in many ways since all the existing tools
>>> assume that if there is something mounted, there will already be an entry
>>> in the directory for the tools (ls, bash) to pick up on.
>>[...]
>>
>>The fact that mount does not require the parent directory to exist is
>>a bug.  It really ought to.
>>
>>I also consider the //<drive-letter>/ system for referring to drives a
>>Bad Idea (tm) since it conflicts with the UNC namespace.  I'm somewhat
>>thinking of removing this functionality and replacing it with
>>something more sensible but this may be too painful for everyone for
>>us to contemplate it seriously.  :-(
>


I don't know about anyone else but I already use UNC names with cygwin32
with no problem.  Now, of course I'm assuming that someone who tries to 
do this has set up their network shares properly so its possible to see
them and use them.  Otherwise, it should work, regardless of what
significance cygwin32 applies to the "//<drive>/" syntax.  Keep in mind
that the UNC convention is NOT as Mikey suggests above (you don't put a 
colon after the drive... as a matter of fact, what you put in the "drive"
spot is determined by what you called your share when you shared it which
doesn't have to be the hokey drive letter at all!)  All this comes for free 
from using the CreateFile() call as far as I can tell.  I've done some tests
in the past that seem to support this position.  Of course, I've only tried
this using NT (4.0)...




Larry Hall                              lhall AT rfk DOT com
RFK Partners, Inc.                      (781) 239-1053
8 Grove Street                          (781) 239-1655 - FAX
Wellesley, MA  02482-7797               http://www.rfk.com

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019