Date: Mon, 20 Jan 2003 23:36:06 +0300 From: "Eli Zaretskii" Sender: halo1 AT zahav DOT net DOT il To: djgpp-workers AT delorie DOT com Message-Id: <4634-Mon20Jan2003233606+0200-eliz@is.elta.co.il> X-Mailer: emacs 21.3.50 (via feedmail 8 I) and Blat ver 1.8.9 In-reply-to: <3E2C5FEF.E9AC8D1C@phekda.freeserve.co.uk> (message from Richard Dawe on Mon, 20 Jan 2003 20:45:35 +0000) Subject: Re: fstat, fd_props and inventing inodes [PATCH] References: <2593-Wed15Jan2003095337+0200-eliz AT is DOT elta DOT co DOT il> <3E2BFD47 DOT F2FCA284 AT phekda DOT freeserve DOT co DOT uk> <2427-Mon20Jan2003213728+0200-eliz AT is DOT elta DOT co DOT il> <3E2C5FEF DOT E9AC8D1C AT phekda DOT freeserve DOT co DOT uk> Reply-To: djgpp-workers AT delorie DOT com Errors-To: nobody AT delorie DOT com X-Mailing-List: djgpp-workers AT delorie DOT com X-Unsubscribes-To: listserv AT delorie DOT com Precedence: bulk > Date: Mon, 20 Jan 2003 20:45:35 +0000 > From: Richard Dawe > > I tried to think of reasons why _truename was preferred over _fixpath. I don't remember; perhaps the archives of this list will help. > AFAICS > _truename handles subst'd/assign'd/join'd drives, but _fixpath doesn't. That > seems like a good reason for keeping _truename and handling UNCs specially. Yes, I agree. > How do you think we should handle UNCs? > > Should we just ignore UNCs? > > Or should we scan through the list of shares that we have mapped, to determine > the drive letter? The latter, IMHO. Ignoring is no good, since it will break `fstat' for remote files. > (I was assuming that _truename would return the drive letter > for a UNC with a mapped drive.) I think `_truename' cannot do that because that would slow it down: to do the job, you need to loop through 32 possible drives, making a system call for each one of them. So perhaps we should only do that lazily, and/or maybe make the st_dev member be one of the expensive ones which is only computed if a corresponding bit in `_djstat_flags' is cleared.