From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 29434 invoked by alias); 20 Apr 2009 21:36:19 -0000 Received: (qmail 29422 invoked by uid 22791); 20 Apr 2009 21:36:17 -0000 X-SWARE-Spam-Status: No, hits=-2.4 required=5.0 tests=AWL,BAYES_00 X-Spam-Check-By: sourceware.org Received: from rock.gnat.com (HELO rock.gnat.com) (205.232.38.15) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Mon, 20 Apr 2009 21:36:10 +0000 Received: from localhost (localhost.localdomain [127.0.0.1]) by filtered-rock.gnat.com (Postfix) with ESMTP id 034282BAAE2; Mon, 20 Apr 2009 17:36:09 -0400 (EDT) Received: from rock.gnat.com ([127.0.0.1]) by localhost (rock.gnat.com [127.0.0.1]) (amavisd-new, port 10024) with LMTP id WZfDB4FKizy9; Mon, 20 Apr 2009 17:36:08 -0400 (EDT) Received: from joel.gnat.com (localhost.localdomain [127.0.0.1]) by rock.gnat.com (Postfix) with ESMTP id C20C92BAAE1; Mon, 20 Apr 2009 17:36:08 -0400 (EDT) Received: by joel.gnat.com (Postfix, from userid 1000) id 455BBF58C9; Mon, 20 Apr 2009 14:35:59 -0700 (PDT) Date: Mon, 20 Apr 2009 22:34:00 -0000 From: Joel Brobecker To: Tom Tromey Cc: gdb@sourceware.org Subject: Re: prefer in-tree libiconv over the system libiconv? Message-ID: <20090420213559.GS2904@adacore.com> References: <20090420183619.GB5858@adacore.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.18 (2008-05-17) Mailing-List: contact gdb-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-owner@sourceware.org X-SW-Source: 2009-04/txt/msg00159.txt.bz2 > I think the relevant scenario is something like the old Cygnus "devo" > tree, where you would have libiconv always present and also use the > single soruce tree on all hosts. In this situation, would you prefer > to pick up the system iconv on Linux? Or would you prefer to always > use libiconv and have identical behavior across hosts (at some cost; I > think the native glibc iconv probably handles more charsets). My very personal preference is to go for consistency. I like to know that the library will behave the same from HP/UX to GNU/Linux. Also, I don't like depending on system libraries, because I lose a bit of control over what the user ends up using. I am afraid of situations where we cannot reproduce and identify a bug simply because the user has a different version of the library from the version that I have. I think that the current approach would have made more sense if libiconv was in fact made part of the GDB sources, thus forcing people to have the libiconv sources be part of their tree. That being said, now wearing my AdaCore hat, I'm perfectly happy to have to use a command-line switch. We already do that with libexpat, and it's done in a script anyway... I'm actually happy to work on a patch once others have had a chance to tell us what they would prefer. If no one speaks up, then I guess I'll implement what my prefered alternative (this is my way of threatening everyone into participating ;-). -- Joel