From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 4.0.0 (2022-12-14) on finch.gentoo.org X-Spam-Level: X-Spam-Status: No, score=0.3 required=5.0 tests=DMARC_NONE,MAILING_LIST_MULTI, RDNS_DYNAMIC autolearn=no autolearn_force=no version=4.0.0 Received: from mail.thpoon.com (CPE0080c8f2c614.cpe.net.cable.rogers.com [24.42.106.79]) by chiba.3jane.net (Postfix) with SMTP id 9D8182019AB0 for ; Sat, 23 Mar 2002 04:28:14 -0600 (CST) Received: (qmail 83328 invoked from network); 23 Mar 2002 10:24:13 -0000 Received: from unknown (HELO tea.thpoon.com) (192.168.1.2) by 192.168.1.1 with SMTP; 23 Mar 2002 10:24:13 -0000 Received: (qmail 18153 invoked by uid 1000); 23 Mar 2002 10:29:11 -0000 To: gentoo-dev@gentoo.org X-Face: 0=A/O5-+sE[Tf%X>rYr?Y5LD4,:^'jaJ!4jC&UR*ZrrK2>^`g22Qeb]!:d;}2YJ|Hq"LHdF OX`jWX|AT-WVFQ(TPhFVak)0nt$aEdlOq=1~D,:\z5QlVOrZ2(H,mKg=Xr|'VlHA="r Mail-Reply-To: agenkin-dated-1018088102.c2adef@thpoon.com Date: Sat, 23 Mar 2002 05:29:10 -0500 Message-ID: <87vgbnshl5.fsf@tea.thpoon.com> User-Agent: Gnus/5.090006 (Oort Gnus v0.06) XEmacs/21.4 (Common Lisp, i686-pc-linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii From: Arcady Genkin X-Delivery-Agent: TMDA/0.49 (Python 2.2 on linux2) Subject: [gentoo-dev] NVIDIA and /usr/lib/OpenGL.so thoughts Sender: gentoo-dev-admin@gentoo.org Errors-To: gentoo-dev-admin@gentoo.org X-BeenThere: gentoo-dev@gentoo.org X-Mailman-Version: 2.0.6 Precedence: bulk Reply-To: gentoo-dev@gentoo.org X-Reply-To: agenkin-dated-1018088102.c2adef@thpoon.com List-Help: List-Post: List-Subscribe: , List-Id: Gentoo Linux developer list List-Unsubscribe: , List-Archive: X-Archives-Salt: 0b8a260c-5bb3-4401-abfb-3f77cdf567f1 X-Archives-Hash: 657c4b23ea7dc85f4b2c15b4e2dd5576 Since we updated to the new NVIDIA drivers layout I have seen a lot of noise about it in -users and in the bug tracker. I still think that we did the right thing by moving the drivers into /opt, and I still think that the applications that don't honor run-time linker's resolution of libGL.so to /opt/nvidia/lib are broken. However, I think that we can improve the situation a bit. Some premises: 1. Drobbins voice a hypothesis that some programs (quake) have "/usr/lib/libGL.so" hardcoded and use a system call to load the lbrary. 2. It seems that compilation sybsystems of other programs (gltron, perhaps?) define run-time library paths at compile time by passing the linker "-rpath" option. 3. Xfree installs *symlinks* /usr/lib/libGL.so -> /usr/X11R6/lib/libGL.so /usr/lib/libGL.so.1 -> /usr/X11R6/lib/libGL.so.1 it's important that they are symlinks. A possible solution: 1. Xfree ebuild does *not* install the symlinks. 2. Instead we create a separate package that would install just the symlinks into /usr/lib (for people that do not use NVIDIA drivers). We can name it "xfree-gl-symlinks" or just "xfree-gl". 3. For the people who use NVIDIA drivers, we make nvidia-glx pacakge install the synlinks in /usr/lib, pointing to /opt/nvidia/lib, naturally. This ain't pretty, and it's a kludge. I don't like it much, but I think that this will make lives of many users easier, at a nominal cost. If anyone can think of another way to fix the problem, I hope that a better idea exits. ;^) Comments? -- Arcady Genkin Don't read everything you believe.