Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT sources DOT redhat DOT com Delivered-To: mailing list cygwin AT sources DOT redhat DOT com Message-ID: <3974026E.B11D25BE@murdoch.edu.au> Date: Tue, 18 Jul 2000 15:08:30 +0800 From: Stewart Greenhill X-Mailer: Mozilla 4.72 [en] (WinNT; I) X-Accept-Language: en MIME-Version: 1.0 To: cygwin AT sourceware DOT cygnus DOT com Subject: Re: Have rename() semantics changed? Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit > At 12:23 PM 7/13/2000, Chris Faylor wrote: > >On Thu, Jul 13, 2000 at 05:15:06PM +0100, Fifer, Eric wrote: > > > > > >Stewart Greenhill wrote: > > >> The latest cygwin seems not to allow open files to be renamed. This > > >> works under Unix. It is not normally allowed under Windows, but worked > > >> in previous versions of the cygwin environment. > > >> > > >> Is this an official change, or could it be a bug? > > > > > >I compared strace's on b20.1 and 1.1.2 and the shared argument to > > >CreateFileA has changed. FILE_SHARE_DELETE was dropped from > > >host_dependent.shared in dcrt0.cc (host_dependent_constants::init). > > >This was the comment at the time: > > > > > > Sat Mar 18 01:32:04 2000 Christopher Faylor > > > > > > * dcrt0.cc (host_dependent_constants::init): Eliminate DELETE flag > > > from shared constant. > > > > > >I put it back and your test program works again. Perhaps Chris can > > >shed some light on the reason for the change. > > > >Can you still unlink opened files on Windows 9x, Windows NT, and W2K > >with this change? I believe that is why I removed this. Either that > >or it prevented files from being opened. Sorry that I don't remember. > > > >If the above still works, then I'll put this back and comment this > >so we won't have to guess next time. > According to the MSDN, FILE_SHARE_DELETE is a NT/W2K thingy. That implies > that this functionality is only available there. One would hope that > enabling it doesn't cause 9x to misbehave but who knows. I suppose the > constant could be conditionally compiled in if need be... Folks, Thanks for looking into this problem. I hope that this can be fixed for the next cygwin release. In the mean time, it looks like I probably have to step back to B20.1. I should have mentioned that I was having this problem with Win NT 4. I haven't tried it under any other version of Windows. Cheers, Stewart -- Want to unsubscribe from this list? Send a message to cygwin-unsubscribe AT sourceware DOT cygnus DOT com