From mboxrd@z Thu Jan 1 00:00:00 1970 From: ludo@gnu.org (Ludovic =?utf-8?Q?Court=C3=A8s?=) Subject: Re: [PATCH 1/4] lint: Add packages with suffix ':bin' to 'inputs-should-be-native'. Date: Fri, 22 Jul 2016 15:32:48 +0200 Message-ID: <878twtpzlr.fsf@gnu.org> References: <20160719164916.6597-1-david@craven.ch> <20160719164916.6597-2-david@craven.ch> <87r3aobnoi.fsf@gnu.org> <87y44vnpq3.fsf@gnu.org> <87poq7kmbb.fsf@gnu.org> <87ad02716d589c2bf486478bc76199e0@tobias.gr> 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]:33748) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bQaZg-0005IJ-Kf for guix-devel@gnu.org; Fri, 22 Jul 2016 09:33:09 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bQaZe-0000Ql-NF for guix-devel@gnu.org; Fri, 22 Jul 2016 09:33:07 -0400 In-Reply-To: <87ad02716d589c2bf486478bc76199e0@tobias.gr> (Tobias Geerinckx-Rice's message of "Thu, 21 Jul 2016 21:27:49 +0200") 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: Tobias Geerinckx-Rice Cc: guix-devel , Guix-devel , David Craven Tobias Geerinckx-Rice skribis: > On 2016-07-21 18:06, ludo@gnu.org wrote: >>>> (inputs `(("somethingsilly" ,glib "bin"))) >> >> This "somethingsilly" is just a label that allows you to refer to an >> input on the build side, in idioms like: >> >> (assoc-ref inputs "somethingsilly") >> =3D> "/gnu/store/=E2=80=A6-glib-42.0-bin" > > I remember writing something like > > (inputs `(("libuuid" ,util-linux))) > > once, to indicate that libuuid really is all we need and that one > could swap out util-linux for another =E2=80=94 extremely hypothetical = =E2=80=94 > libuuid provider. > > Was this, in hindsight, somethingsilly? :-) I think it=E2=80=99s fine, but a comment would work well too. > If so, is there still a reason to keep all this duplication around? Build systems haven=E2=80=99t migrated yet to gexps. There=E2=80=99s a bra= nch to do that, =E2=80=98wip-build-systems-gexp=E2=80=99, but there=E2=80=99s a perfo= rmance regression compared to what we have, which is why I have not merged it. That=E2=80=99ll happen, someday, but even then, I=E2=80=99m not sure we=E2= =80=99ll remove labels; at least we won=E2=80=99t remove them overnight, for sure. ;-) >> This predates g-expressions, which solve this problem more elegantly. > > Interesting! I always assumed[1] gexps were one of the foundations for > Guix since day 0. In an ideal world it would have been this way. :-) Ludo=E2=80=99.