From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mark H Weaver Subject: Re: [PATCH] gnu: gzochi: Update to 0.10. Date: Tue, 23 Aug 2016 01:20:31 -0400 Message-ID: <87a8g45ae8.fsf@netris.org> References: <874m6dsa5l.fsf@netris.org> Mime-Version: 1.0 Content-Type: text/plain Return-path: Received: from eggs.gnu.org ([2001:4830:134:3::10]:58024) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bc4A0-0005Jo-8J for guix-devel@gnu.org; Tue, 23 Aug 2016 01:22:05 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bc49x-0003Mz-24 for guix-devel@gnu.org; Tue, 23 Aug 2016 01:22:04 -0400 Received: from world.peace.net ([50.252.239.5]:45011) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bc49w-0003JY-U3 for guix-devel@gnu.org; Tue, 23 Aug 2016 01:22:00 -0400 In-Reply-To: (Julian Graham's message of "Mon, 22 Aug 2016 08:59:16 -0400") List-Id: "Development of GNU Guix and the GNU System distribution." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-devel-bounces+gcggd-guix-devel=m.gmane.org@gnu.org Sender: "Guix-devel" To: Julian Graham Cc: guix-devel Julian Graham writes: > Hi Mark, > > On Mon, Aug 22, 2016 at 12:24 AM, Mark H Weaver wrote: >> This update seems to have broken gzochi on i686 and armhf. Segfaults >> and aborts occur during the test suite. > > Oh dear, that's embarrassing. > > >> https://hydra.gnu.org/build/1444785 (i686) >> https://hydra.gnu.org/build/1444781 (armhf) >> >> Build logs are available from those pages. As I write this, our >> continuous integration system has not yet attempted the build on >> mips64el, but when it does the results will be available here: >> >> https://hydra.gnu.org/build/1444727 (mips64el) >> >> Any idea what's going wrong here? > > It's almost certainly some sloppy handling of 64-bit types on 32-bit > platforms. (The mips64el build succeeded.) > > Sorry for the inconvenience! No worries, bugs happen. > What are my options for addressing this? > It may take me a few days to get this patched. If you have 'guix' installed, you can run: guix build --system=i686-linux --keep-failed gzochi to reproduce the same build on your machine, and to leave the failed build directory in /tmp/guix-build-* for further investigation. Another approach is to run: guix environment --pure --system=i686-linux gzochi to spawn a subshell with environment variables set to mostly reproduce the build environment used in the "guix build" command above, but outside of the isolated build container. If you can reproduce the problem this way, it should be far more convenient for debugging and for development of a fix. If you can fix the problem on i686, please let me know and I'd be glad to test the same fix on armhf. What do you think? Mark