From mboxrd@z Thu Jan 1 00:00:00 1970 From: ludo@gnu.org (Ludovic =?UTF-8?Q?Court=C3=A8s?=) Subject: bug#26791: Hydra doesn't run make before using Guix Date: Tue, 09 May 2017 23:29:47 +0200 Message-ID: <87mvalwwfo.fsf@gnu.org> References: <87a86qvmvh.fsf@netris.org> <20170507121148.375a0280@scratchpost.org> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Return-path: Received: from eggs.gnu.org ([2001:4830:134:3::10]:57353) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1d8Ci4-0001cA-8C for bug-guix@gnu.org; Tue, 09 May 2017 17:30:22 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1d8Ci2-0001oe-7N for bug-guix@gnu.org; Tue, 09 May 2017 17:30:20 -0400 Received: from debbugs.gnu.org ([208.118.235.43]:60249) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1d8Ci2-0001oB-2i for bug-guix@gnu.org; Tue, 09 May 2017 17:30:18 -0400 Received: from Debian-debbugs by debbugs.gnu.org with local (Exim 4.84_2) (envelope-from ) id 1d8Ci1-00007B-S1 for bug-guix@gnu.org; Tue, 09 May 2017 17:30:17 -0400 Sender: "Debbugs-submit" Resent-Message-ID: In-Reply-To: <20170507121148.375a0280@scratchpost.org> (Danny Milosavljevic's message of "Sun, 7 May 2017 12:11:48 +0200") List-Id: Bug reports for GNU Guix List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: bug-guix-bounces+gcggb-bug-guix=m.gmane.org@gnu.org Sender: "bug-Guix" To: Danny Milosavljevic Cc: 26791@debbugs.gnu.org Hi Danny, Danny Milosavljevic skribis: > For reference, the relevant commits are (newest first) 360874dd1c1aafddd8= 2e389b9da09f69bddb2a20 (not really suspicious), 7e3f7acbe0885fd8a6e509fd0c3= 09d1f4d9d3097, bf8b09169c7d8ae48ac99ea237b82867ee9aa885, 370ae085b5107a0928= fd744a987fb4070bdf6a00, 83071b052a0585f7166245b7ba793436e104f8fe, 0315abe60= 8007f1fed96cdcae21b6a83902b25e6, 9530e73b496fefe65dcb936825b6beda79f7fdf2, = 33f0aa88155ee0718f21e7de24eb539cb9872217, 40fad1c24ce60076e26f6dc8096e4716d= 31d90c3 (!). > > Note that this all works fine after Guix has been built using make, so I = don't think the problem is in the source code of Guix. > > As for that Hydra job: > >>Switched to a new branch 'fetchgit' >>;;; compiled /home/hydra/.cache/guile/ccache/2.0-LE-8-2.0/usr/local/bin/h= ydra-eval-guile-jobs.go >>adding `/gnu/store/vsw3112dw48ai5arsxf1qsvma8y4fvqp-git-export' to the lo= ad path >>;;; compiling /gnu/store/vsw3112dw48ai5arsxf1qsvma8y4fvqp-git-export/buil= d-aux/hydra/../../gnu/packages/xnee.scm > > It seems that hydra auto-compiles some Guix modules: > > So this Hydra job should run make in /gnu/store/vsw3112dw48ai5arsxf1qsvma= 8y4fvqp-git-export first. > > Unfortunately, I have no idea about Hydra - and "grep -r fetchgit ." insi= de "guix" doesn't turn up anything either. Not a bug! :-) Hydra and Cuirass first go through an =E2=80=9Cevaluation=E2=80=9D step, wh= ich consists in producing a list of derivations + meta-data from Guix/GuixSD. Concretely, Hydra runs build-aux/hydra/gnu-system.scm, which returns that list (you can run =E2=80=9Cmake hydra-jobs.scm=E2=80=9D to see that re= sult.) As can be seen at the top of gnu-system.scm, the idea is to let Guile=E2=80= =99s auto-compilation kick in rather than have an explicit build step. It simplifies gnu-system.scm at the expense of providing messy output; it=E2= =80=99s also quite expensive. I=E2=80=99m open to a brainstorming session on this. I=E2=80=99m not sure = how to improve the situation myself. We have a similar problem with =E2=80=98guix pull=E2=80=99 in fact: should we first compile everything and then run, or compile everything but the (gnu packages =E2=80=A6) modules, or=E2=80=A6? Ludo=E2=80=99.