RE: unlink() problem

["Larry Hall (RFK Partners, Inc)" <lhall at rfk dot com>]

At 03:47 PM 8/8/2001, John Wiersba wrote:
>Thanks for your reply, Larry.  Unfortunately, this is a bad bug since the
>simple perl script below removes the original file (whoops!) since it
>assumes that if you have permission to unlink the original file, then you
>have permission to create a new file of the same name (I believe this is
>always true on unix).  The net result is that you lose your data.
>
>I can't think of anyway around it except to know that you can't do inplace
>editing on cygwin.


Right.  I suppose if one could queue up the file creation as well, 
this might help.  Still it ain't UNIX semantics by a long shot!


> > -----Original Message-----
> > From: Larry Hall (RFK Partners, Inc) [mailto:lhall@rfk.com]
> > Sent: Tuesday, July 31, 2001 6:59 PM
> > To: John Wiersba; 'cygwin'
> > Subject: Re: unlink() problem
> > 
> > 
> > At 06:22 PM 7/31/2001, John Wiersba wrote:
> > >I'm having a problem with unlink().  I've scanned the FAQ 
> > and searched some
> > >in the newsgroup archives for information about the 
> > following problem.  I
> > >did find a post (see
> > >http://sources.redhat.com/ml/cygwin/2001-05/msg01095.html) 
> > but no apparent
> > >resolution.
> > >
> > >$ touch asdf
> > >$ perl -i -pe 1 asdf
> > >Can't do inplace edit on asdf: Permission denied.
> > >
> > >What perl is doing is equivalent to:
> > >
> > >$ perl -e '
> > >    open A, "< asdf" or die "one $!";  # open asdf for read
> > >    unlink "asdf";
> > >    open B, "> asdf" or die "two $!";  # open asdf for write
> > >    while (<A>) { print B }            # copy from A to B
> > >'
> > >two Permission denied at -e line 4.
> > >
> > >The problem:  after the unlink(), I can't open the same 
> > filename for write
> > >as I can on unix.  If I pause the program after unlinking 
> > asdf, it is still
> > >there and can be seen by ls, but can't be read by ls -l.  
> > >
> > >Is there a workaround for this behavior?  Is it considered a 
> > bug or is it
> > >just a quirk of the cygwin implementation of trying to implement unix
> > >filesystem semantics on top of windows?
> > 
> > 
> > The latter.  This has been quite a day for this issue!  This is the 
> > third time it came up.  Perhaps this qualifies as an FAQ now! ;-)
> > 
> > See my previous posts about this subject from earlier in the day:
> > 
> > http://www.cygwin.com/ml/cygwin/2001-07/msg01797.html
> > http://www.cygwin.com/ml/cygwin/2001-07/msg01774.html
> > 
> > There's also a related FAQ, although the message it mentions doesn't
> > seem to pop up anymore, which gives insight into the underlying 
> > implementation.  See:
> > 
> > http://www.cygwin.com/faq/faq_4.html#SEC57
> > 
> > Of course, the code is the "last word" on this subject in terms of the
> > actual implementation.
> > 
> > 
> > 
> > 
> > 
> > 
> > 
> > Larry Hall                              lhall@rfk.com
> > RFK Partners, Inc.                      http://www.rfk.com
> > 118 Washington Street                   (508) 893-9779 - RFK Office
> > Holliston, MA 01746                     (508) 893-9889 - FAX
> > 


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/