From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 1109 invoked by alias); 22 Nov 2004 14:05:46 -0000 Mailing-List: contact gdb-patches-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sources.redhat.com Received: (qmail 1062 invoked from network); 22 Nov 2004 14:05:37 -0000 Received: from unknown (HELO relay3.mail.uk.clara.net) (80.168.70.143) by sourceware.org with SMTP; 22 Nov 2004 14:05:37 -0000 Received: from adsl-2-solo-172-54.claranet.co.uk ([80.168.172.54] helo=[172.31.0.98]) by relay3.mail.uk.clara.net with esmtp (Exim 4.34) id 1CWEow-000GvT-F5; Mon, 22 Nov 2004 14:05:35 +0000 Message-ID: <41A1F337.8060403@redhat.com> Date: Mon, 22 Nov 2004 14:05:00 -0000 From: Nick Clifton User-Agent: Mozilla Thunderbird 0.7.3 (X11/20040803) MIME-Version: 1.0 To: Richard Earnshaw CC: Richard Earnshaw , binutils@sources.redhat.com, gdb-patches@sources.redhat.com, newlib@sources.redhat.com Subject: Re: Dejagnu: use -isystem to include system header files. References: <89A528FE6DB0FA44877BB2F05B8467180160F913@ZIPPY.Emea.Arm.com> In-Reply-To: <89A528FE6DB0FA44877BB2F05B8467180160F913@ZIPPY.Emea.Arm.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-SW-Source: 2004-11/txt/msg00436.txt.bz2 Hi Richard, > The issue wasn't really about picking up gcc/include/limits.h over > newlib/include/limits.h, it was about how we processed the newlib > version when in strict ANSI mode. That should be handled correctly when > we use -isystem (because that relaxes the rules). So I think we should > try and get the search order back to what it was before, but when still > using -isystem. I am not sure that this is possible. The /gcc/include directory is added via an -isystem switch that is synthesised by the gcc (or xgcc) compiler driver. It is processed before any-isystem switch specified by the user on the command line so that gcc's fixed versions of system header files will be included before the systems. Thus if the dejagnu test harness uses the -isystem switch to add the newlib/libc/include directory it is always going to be after /gcc/include in the list of directories searched. This sounds correct to me. The intention is that the fixed version of limits.h should be included before the system version of limits.h. Thus builtins-config.h should not assume that #include will return the system limits.h. It might well return a fixed version instead. Cheers Nick