From mboxrd@z Thu Jan 1 00:00:00 1970 From: ludo@gnu.org (Ludovic =?utf-8?Q?Court=C3=A8s?=) Subject: Re: [PATCH] gnu: gnutls: Replace with 3.5.8 [fixes GNUTLS-SA-2017-{1, 2}]. Date: Thu, 12 Jan 2017 15:18:24 +0100 Message-ID: <8760lk4b0v.fsf@gnu.org> References: <20170110181803.GB28153@jasmine> <20170110221532.3691-1-ludo@gnu.org> <20170110223626.GC8431@jasmine> <87inpmy1uj.fsf@gnu.org> <20170110230839.GA10774@jasmine> <20170111171741.GA21549@jasmine> 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]:41402) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1cRgD2-0006pY-P4 for guix-devel@gnu.org; Thu, 12 Jan 2017 09:18:37 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1cRgCx-0002GT-BR for guix-devel@gnu.org; Thu, 12 Jan 2017 09:18:32 -0500 In-Reply-To: <20170111171741.GA21549@jasmine> (Leo Famulari's message of "Wed, 11 Jan 2017 12:17:41 -0500") 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: Leo Famulari Cc: guix-devel@gnu.org Leo Famulari skribis: > On Tue, Jan 10, 2017 at 06:08:39PM -0500, Leo Famulari wrote: >> On Tue, Jan 10, 2017 at 11:41:40PM +0100, Ludovic Court=C3=A8s wrote: >> > Leo Famulari skribis: >> > > The last time I noticed an exported replacement, I found that it was >> > > resolved non-deterministically (using `guix build -S` in a loop), and >> > > we made the replacement private: >> > > >> > > http://git.savannah.gnu.org/cgit/guix.git/commit/?id=3D69aa6e0995b55= a38d5ce174602a107645be726d5 >> >=20 >> > I can=E2=80=99t seem to reproduce it here, but it sounds like an inter= esting >> > bug. :-) >>=20 >> Me neither. IIRC, it would manifest at least 1/10 iterations of the >> loop. > > I can reproduce it by applying the diff below on > 88f2dd1ddf8123f628ee0b64406b6fd2a6a9f076 and running: > > $ while true; do ./pre-inst-env guix build gnutls -S; done > guix build: warning: ambiguous package specification `gnutls' > guix build: warning: choosing gnutls-3.5.4 from gnu/packages/tls.scm:139:2 > /gnu/store/kfjzkvlxjp98mcf356ay027vy5p5lsp3-gnutls-3.5.8.tar.xz > guix build: warning: ambiguous package specification `gnutls' > guix build: warning: choosing gnutls-3.5.4 from gnu/packages/tls.scm:216:2 > /gnu/store/kfjzkvlxjp98mcf356ay027vy5p5lsp3-gnutls-3.5.8.tar.xz > guix build: warning: ambiguous package specification `gnutls' > guix build: warning: choosing gnutls-3.5.4 from gnu/packages/tls.scm:216:2 > /gnu/store/kfjzkvlxjp98mcf356ay027vy5p5lsp3-gnutls-3.5.8.tar.xz > guix build: warning: ambiguous package specification `gnutls' > guix build: warning: choosing gnutls-3.5.4 from gnu/packages/tls.scm:139:2 > /gnu/store/kfjzkvlxjp98mcf356ay027vy5p5lsp3-gnutls-3.5.8.tar.xz Oh, I see. I can imagine that this has to do with the ordering of package objects and first-class variables in the various hash tables and vhashes involved. I=E2=80=99d say it doesn=E2=80=99t matter much because we shouldn=E2=80=99t= be providing two packages with the exact same name and version in the first place (which is what this patch does). Ludo=E2=80=99.