X-Authentication-Warning: delorie.com: mailnull set sender to djgpp-workers-bounces using -f Date: Thu, 9 May 2002 09:54:55 -0700 From: Zack Weinberg To: pavenis AT lanet DOT lv Cc: Andris Pavenis , djgpp-workers AT delorie DOT com, Eli Zaretskii , Nick Clifton , gcc-patches AT gcc DOT gnu DOT org Subject: Re: RFA: Ignore DOS end-of-line characters (ctrl-Z) unless -W Message-ID: <20020509165455.GA3910@codesourcery.com> References: <3CDA764D DOT 1410 DOT A1F3EA AT localhost> <3CDAC9D0 DOT 15538 DOT 1E838FC AT localhost> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <3CDAC9D0.15538.1E838FC@localhost> User-Agent: Mutt/1.3.28i 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 On Thu, May 09, 2002 at 07:11:12PM +0300, pavenis AT lanet DOT lv wrote: > On 9 May 2002 at 8:42, Zack Weinberg wrote: > > > On Thu, May 09, 2002 at 01:14:53PM +0300, pavenis AT lanet DOT lv wrote: > > > > > > OK. Here it is (tested for DJGPP with 3.1 branch) > > > > O_TEXT, not O_BINARY. Also, based on other discussion with Eli, the > > check should apply to all file descriptors, not just stdin, so please > > move it below the call to open. > > > > I don't think so. > > With O_TEXT we will run into another trouble as length of data in memory will > be smaller than one in file. As result read_include_file() will complain that > input file is shorter than expected. > > So unless we want to remove that check, O_BINARY must stay there. > > Current patch changes mode to O_BINARY in rather specific case when > one is feeding source in gcc from pipe. For example > cat foo.c | gcc -c -x c -o foo.o - > In this case file is opened in text mode by default, and I'm getting > a warning I mentioned above. Sorry, I misread !isatty as isatty. This is fine, then. zw